Feasting on My Enemy: Images of Violence and Change in the New Guinea Highlands

Abstract

This article considers changing images and practices of violence seen through the lens of two categories of the imaginary: assault sorcery and cannibalistic witchcraft. The ethnographic and historical locus for this discussion is the Highlands of Papua New Guinea, within which we further concentrate on three ethnographic cases where we have conducted fieldwork: in Hagen, in Pangia, and among the Duna speakers of Lake Kopiago (Figure 1). The time period examined is from the mid-1960s to 1998, coinciding with colonial and postcolonial changes in this region. With colonial pacification and postcolonial transformations in the political economy, many alterations have occurred in indigenous perceptions with regard to violent conflict and to the political values attached to social space. These changes have been reflected in changing notions about sorcery and witchcraft, two sorts of “mystical violence” that have tended to flourish in contexts where open warfare between groups or open interpersonal violence are suppressed or restrained.

We direct our attention to both sorcery and witchcraft because we see these as two overlapping conceptualizations of hidden violence that transgress boundaries of the human body and the body politic, as argued by Mary Douglas and others (Douglas 1966; Lock and Scheper-Hughes 1987; Stewart and Strathern 1997, see also Strathern 1996). Such forms of violence can clearly be seen as alternatives to warfare, or as forms of supernatural warfare themselves. But sorcery and witchcraft accusations and ideas can also be precipitates of patterns of disease and the sickness and death caused by disease. Images connected with sorcery and witchcraft therefore tend to co-vary with changing historical patterns of disease in epidemiological terms. In colonial circumstances these patterns further correlate with the spatial movements of people between communities as well as with the spread of infectious diseases. Epidemiological patterns are therefore a part of the evidence brought forth in this article on why sorcery and witchcraft ideas change over time.

The New Guinea Highlands is an opportune area in which to examine these themes because the history of colonial contact and postcolonial change is relatively recent. Ethnohistorical accounts by the people are an important part of the narrative as a whole. Furthermore, this is a region in which warfare and acts of violence flourished in precolonial times and have subsequently reemerged as a part of alternating patterns of hostility, revenge, and peacemaking between groups. Exchanges of wealth goods are made to make alliances and to pay for killings, but these also are counterbalanced against perceived continuing, if covert, acts of hostility by sorcery and witchcraft. The imagery of cannibalism is used by some groups (in Hagen and among the Duna) to express aspects of aggressive hostility and the transcendence of these aspects by “proper” forms of consumption and exchange. Assault sorcery depends on magical notions of bodily invasion and destruction, which parallel those of witchcraft but without cannibalistic images. The common theme linking these two phenomena is that of bodily invasion, resulting either in destruction of or in consumption of body parts themselves. In both cases such assaults on the body reflect the fact that the victim is regarded as an enemy to be destroyed, hence our theme of “feasting on my enemy.” Legitimate exchange is then seen as substituting the consumption of wealth for the consumption or destruction of bodies, thus resulting in “feasting with my enemy.”

Alterations in perceptions of space over time in these societies can also be related to changes in the political economy. The integration of previously autonomous local polities into administrative areas in colonial time had complex results, causing forms of interdependency accompanied by latent intergroup hostilities. Changes in patterns of production, consumption, and exchange resulting from cash cropping in certain areas (e.g., in Hagen) have also led to new patterns of fears of witchcraft; while alterations in the conditions of communication between local and ethnic groups have led to increased fears of assault sorcery (e.g., in Pangia and among the Duna). Gender relations have also been affected by both economic and political changes, leading in some instances to intensified or renewed fears of female witchcraft (among the Duna).

Our framework involves the idea of political space, or the fields of power that are created and altered in colonial and postcolonial conditions. The power of the state, seen most broadly, has altered political space in the societies we consider, not simply by pacification—the removal of warfare and the creation of peace between groups—but rather by setting up new forms and categories of space within which political power was both denied and exercised. The removal of warfare did not mean the removal of animosities, or the preemption of future hostilities; rather, it provided a basis for their elaboration, with permutations of scale, setting up new conditions for fears of sorcery and witchcraft to operate within.

We begin with a brief review of the theories of hostile sorcery and witchcraft in the contexts of Africa and Melanesia, partly because we believe that our Melanesia-based arguments probably also apply to the materials on change in patterns of witchcraft and sorcery ideas in colonial and postcolonial Africa.

Sorcery, Witchcraft, and Conflict

Images of bodily invasion clearly go with perceptions of conflict between persons, the penetration and permeability of boundaries of the body, personhood, and groups, as stressed in the work of Douglas (1966, 1970, 1991, 1996) and others (e.g., Lewis 1970, 1989; Purkiss 1996: 120). In the earlier literature on sorcery and witchcraft, arising to a fair extent from Africanist ethnography, stress tended to be laid on the basis of such practices in conflict, tension, and strain between persons in positions of competition, jealousy, or incompatibility (e.g., Max G. Marwick’s [1952, 1970] theory of sorcery as a social “strain-gauge” and the notion that sorcery provided an avenue of redress for perceived wrongdoings). It is clear that such a theme has empirical applicability to the contexts this article discusses, where the images of conflict are expressed by the human body and by analogy, the bounded community as the chief site for their symbolic work.

We are primarily interested in contexts of historical change rather than in synchronic typologies of systems. We are interested in comparing processes (social and historical) rather than in setting up definitional typologies. Therefore, in discussing assault sorcery and cannibalistic witchcraft, we do not stress the distinction between the terms sorcery and witchcraft as such; rather, we use these terms for convenience to refer to symbolic syndromes of mystical violence¹ that are partly distinct and partly overlap. In reviewing the question of definitional distinctions between sorcery and witchcraft, several authors have concluded that hard and fast distinctions can be made in hypothetical terms but then fail to fit the complexities of ethnographic fact. Bruce M. Knauft (1985: 346), for example, comments that “opposed definitional terms such as ‘sorcery’ or ‘witchcraft’ leave problematic how intermediate cases are to be labeled and discussed.” He goes on to say that the analysis of how sickness-sending beliefs can articulate with political and symbolic actions is worthwhile, but “pinning these processes to a binary typology may unwittingly obscure some of the richer inter-relationships” (ibid.).

The dimensions in terms of which definitional distinctions have been attempted include the innate versus the learned, involuntary versus voluntary action, and illegitimate versus legitimate power (see, e.g., Ellen 1993: 6). The first two distinctions derive in part from the classic work of E. E. Evans-Pritchard (1976 [1937]: on 11) the Azande, although it is clear from this work that the Zande witch was thought of as sending out “the soul of witchcraft” from his body and therefore as performing an intentional act, even though an accused witch might also claim “that he is unconscious of injuring anyone” (ibid.: 42). Zande witchcraft substance (mbisimo mangu) was thought of as inherited in the body and therefore as involuntarily obtained and innate in the person, yet its exercise was clearly seen as intentional, even if the witch was driven by his emotions of jealousy or dislike. A witch’s actions might thus be seen as a complex intermeshing of the voluntary with the involuntary. It is worthwhile noting here that Evans-Pritchard himself was not interested in ideal type definitions; he merely wished to explain what the Azande meant by the terms mangu (which he glosses as “witchcraft”), soroka (oracles), ngua (magic, characterized by ritual action and a spell, whereas mangu involved no use of spells), and sorcery (bad magic [gbegbere ngua]). His definitions are thus glosses of convenience, used to tie in with Zande expression, rather than being intended as universal classifications. The legitimate versus the illegitimate distinction, as proposed by Michele Stephen (1987: 264), runs into the difficulty, remarked on also by Knauft (1985: 346), that “many Melanesian societies accord some degree of deference to alleged sickness-senders,” while they also “actively disparage or castigate these person.” Stephen herself wished to contrast the powerful sorcerer to whom deaths may be attributed with the despised witch who is accused and punished; this distinction applies in some cases, though not always. The upshot is that we, like Evans-Pritchard, are using terms of convenience, but our aim is to situate the discussion in terms of historical changes, as Malcolm D. McCleod does in his 1972 analysis of the history of witchcraft among the Azande.

In Africanist studies of the historical contexts involved in changing patterns of witchcraft accusations, one point that is made very clear is that the societies concerned had been subjected on the whole to severe political and economic dislocations—through wage labor, missionization, and the growing dissolution of social bonds resulting from the movements of people and the challenges to moral codes and patterns of local authority. Douglas (1991: 727) notes that in Malawi, Zaire, Zambia, and Zimbabwe, where many ethnographic studies were done, “the old political systems were not functioning” and witchcraft accusations between rivals for village power therefore arose in contexts where other ways of obtaining redress against an unpopular leader or of contesting succession to an office of leadership were no longer available. Although in an earlier work Douglas (1970: xx) points out that propositions about “social breakdown” and “increase in frequency of accusation” are usually empirically untestable, we are on safer ground in pointing out the demographic, spatial, and structural changes themselves. In situations of increased movements of people, it is thus likely that arenas of ambiguity and distrust in social relations—cited as those in which witchcraft accusations are made—will also increase (referred to in ibid.: xvii). The developmental cycle of competition for local power in a village will also be exacerbated by wider processes of social instability and so give rise to witchcraft accusation, such as John Middleton (1987: 137, 153) notes for the Lugbara of the Sudan. The same holds for sorcery and witchcraft in other parts of the world (on South-East Asia, see Ellen 1993: 16). It is not simply a matter of arguing for an increased frequency of types of action; rather, it is a matter of studying the changing loci of accusations and ways of handling these over time and of relating these changes to wider processes. In this regard the conclusions from African studies certainly apply, mutatis mutandis, to some of the Melanesian cases this article considers (e.g., the Duna). In a section on social change in her survey on sorcery and witchcraft in Melanesia, Stephen (1987: 277–88) argues similarly, pointing out that her legitimate versus illegitimate distinction is intended to apply to structures before colonial changes. The difficulty here lies in supposing that the structures in place before colonialism were necessarily stable rather than fluid and changing. But where changes are recent and documented, we can make some headway in delineating structural transformations: for example, when sorcery, like wealth, is no longer restricted to an elite but is attributed to the populace at large. ²

In one of her surveys of different patterns of boundary-maintaining -redefining behavior, Douglas (1970: xxvi–xxvii) points out that a witch may be seen either as an outsider or as an internal enemy (a member of a rival faction or having outside liaisons, or as a dangerous deviant). These patterns may change over time, but in one of our categories, the assault sorcerer, the predominant assertion is that the perpetrator is an outsider, a classic version of a terrifying enemy.

Assault Sorcery and Images of Violence

Assault sorcery is generally correlated with distance and hostility. ³ The sorcerer is seen as an outsider who penetrates a community area or isolates a victim on the periphery of such an area, in a garden or a secluded pathway, and makes a physically aggressive attack by stunning the victim, butchering the victim’s internal organs and removing them, sewing the person up again and sending him or her home to eventually die. Often, if not usually, the image is that assault sorcerers operate in squads that are trained by adepts. Assault sorcery is closely cognate with notions of warfare and is therefore likely to flourish when community spaces defined by warfare have lost their definition because of pacification and colonial restructuring, when there is a struggle to define new spaces and exercise power within them.

Furthermore, in assault sorcery the sorcerer directly confronts and overcomes the victim by minatory force. In a classic examination of types of sorcery and violence among the Fore of Eastern Highlands Province in Papua New Guinea, Jate, Kamano, and Usurufa, R. M. Berndt (1962: 224) explicitly notes that in performing this kind of sorcery “the sorcerers have two main intentions in mind. They have received an injury which demands retaliatory action . . . . [and] they want to avoid open warfare.” An assault sorcerer who is caught will be shot and his corpse dishonored and abused before being returned to its home village, in the same way as might happen to a war casualty. Fears of such sorcery may rise after the forcible ending of open warfare. In this way sorcery can be regarded as a form of feuding between groups. Nevertheless, it is a special kind of feuding, thought to depend on the exercise of magical powers, and therefore contributes to a higher level of terror between groups, making “pacification” less “peaceful” than it might otherwise be.

In his wide-ranging survey of types of “payback” activities in Melanesia, G. W. Trompf (1994: 64) makes a similar point when he notes that an attribution of deaths to sorcerers or witches “is actually a common post-contact development in many areas of Papua and one consequent upon the fact that indiscriminate payback killings or tribal wars have been debarred.” Trompf goes on to discuss several cases in which sorcery accusations have altered with social changes: for example, overcrowding and population pressure in Kalauna (ibid.: 67, citing Young 1971), disease patterns, intergroup mobility, and altered political configurations as among the Mekeo, where circa 1880–1940 the colonial power congregated people in large villages and kept chiefs under tight control so that sorcerers “now being highly mobile and bent on freely negotiating with each other in coteries . . . emerged as a fearful power bloc, accumulating real local political power as steadily as chiefly authority was sucked under expatriate control” (Trompf 1994: 77). Sorcerers had previously acted to inflict punishment and retribution on people at the behest of chiefs. Now they began to act for themselves. Sorcery thus became more than a way of pursuing feud or punishing wrongdoers: it became a mode of politics in itself.

Figure 1.

Locations of major groups in Papua New Guinea discussed in the text.

To pursue our own exposition of these themes, we provide composite ethnographic accounts pertaining primarily to three areas of the Papua New Guinea Highlands: Pangia, Hagen, and the Duna areas, specifically Duna speakers who live near the Strickland River in the Aluni Valley (see Figure 1).

Pangia: Assault Sorcery and Change

The Pangia district in Southern Highlands Province of Papua New Guinea was first brought under administrative control by Australian patrol officers in the 1950s. It contains some twenty thousand speakers of Wiru, horticulturalists and pig-rearers, who were engaged in classic patterns of warfare and exchange between local groups until the time of pacification and missionization. In the 1960s in this area there were some twenty partly colonially created, named villages, containing various groups and group segments, whose members saw themselves in certain contexts as communities. At the southernmost part of the area, an eight-hours walk from the next village, lay Tangupane, a place with fewer than one hundred residents, a high proportion of unmarried men, a low altitude, and an immense reputation for being the home of feared assault sorcerers, called mãua or uro (these names would be pronounced only furtively by informants). People were thought to be at great risk from these sorcerers if they traveled to Tangupane and especially if they slept there. Visits to the outside latrine at night could not be made alone. One person had to stand guard at the latrine’s entrance to watch for a sorcerer’s approach, while another watched from the house door. But the uro supposedly did not venture far beyond Tangupane and its environs.

By the 1980s residents of the Pangia district believed that the number of these sorcerers had greatly multiplied and that they now ranged over the entire Wiru area, even besetting people near the government station. Why this change in perception? We suggest that colonial power had drained the villages of some of their indigenous powers of setting boundaries and set up a space in which invisible, uncontrollable powers that previously were held into local spaces by structures of warfare and restrictions of movement were now seen as moving more swiftly and widely. The space seen as that of pacification and development by administrators was seen by the people as the space of these invisible powers let loose from their previous local contexts.

The colonially created villages were material concentrations of more dispersed forms of sociality that had existed previously. The villages, which were themselves beset by internal factionalism, became bounded arenas of social relations outside of which space was seen as less positively social. Pigs were no longer allowed to be kept in village houses; they were fed and stalled in small huts in garden areas defined as in some way “wild.” The creation of these wild areas set the stage for fears of sorcerers threatening the boundaries of the village settlements, and internal conflicts in the villages made people nervous and apprehensive of hostility. The previous solutions of social spacing were for the time being no longer available.

This example shows that in restructuring community spaces the colonial power in Papua New Guinea set up new sorts of local units and in doing so produced problems of how to define community contexts. Forces of sorcery in Pangia, seen as threatening the integrity of local places, were also seen as forms of bodily invasions. The assault sorcerer lurking on the village edge expresses this dual imagery.

There was no known counteraction against mãua, hence the great fear it inspired. Mãua sorcerers presented an extra threat to women, since the sorcerers were said to rape women before killing them. The sorcerer’s presence was marked by heat that suffocated the victim. Upon sending the victim home, the sorcerer first asked, “Where does the sun rise/set?” If the victim reversed the proper answer, he was satisfied and sent the victim off, with his or her own kidney in the mouth. Upon arrival at home, the victim began to roast and eat his or her own kidney in an autocannibalistic act and was then recognized by kin as doomed to die. Here we see the victim in a state of disorientation, being no longer aware of what is inside or outside or how to relate himself or herself temporally or spatially to the surroundings. This is similar to the type of confusion experienced by persons who are placed in particularly terrifying situations; for example, a soldier who does not realize that an arm or a leg has been blown off during a battle until there is a lull in the fighting and the immediate terror has lessened so that he can take note of his own body.

Among the Daribi, neighbors of the Wiru to the southeast, assault sorcery was known as kebidibidi. Apparently, there were methods of divination to determine who had caused a death by it, as well as statements that counterraiding parties would be organized in revenge (Wagner 1967: 52). ⁴ Roy Wagner (ibid.: 53) points out that a larger number of deaths was attributed to this kind of sorcery than could be feasible. Assault sorcery thus played a special part in the Daribi imaginary economy of death. It would be particularly interesting to know if this situation had recently emerged (i.e., since “pacification”), as Wagner notes that most of the adult deaths that took place during his first fieldwork period were attributed to assault sorcery (ibid.). This was during 1963–65, ten years after the beginning of Australian colonial administration in the area. Possibly this represented a shift in consciousness of senses of agency resulting from “pacification,” as we have suggested for the Wiru. The sense of control of boundaries that collapsed inward during colonial times may have driven communities such as these to attempt to reestablish new ways of explaining what was occurring around them and differentiating themselves from the potentially hostile “others” surrounding them, and increasingly among them, whom they were not allowed to kill in warfare. The inability to express collective group agency through warfare may thus have increased rumors of assault sorcery attacks in both areas.

Witchcraft and Social Epidemiology among the Melpa

Melpa speakers of Mount Hagen in Western Highlands Province of Papua New Guinea number some eighty thousand people, divided into tribes and clans, intensively linked by exchanges and by intergroup hostilities belonging both to the past and to contemporary contexts. The Melpa, who have an elaborate array of ideas about the causation of sickness and death, do not appear to have the idea of assault sorcery in quite the way the Daribi and the Wiru do. They have notions about kum, however, a kind of witchcraft that is believed to operate in a terrifying way reminiscent of assault sorcery, since the kum may eat the inner parts of a person, causing its host to become a cannibal seeking out corpses to consume. The technology of death that eventuates from either assault sorcery or witchcraft is thus quite similar. It is pertinent, therefore, to compare the Melpa’s ideas of kum with assault sorcery in Pangia, since both destroy those inner parts that show the presence of life force and human feelings.

The core idea in notions of witchcraft among the Melpa is greed (Strathern 1982; Stewart and Strathern 1997; cf. Kahn 1986 on Wamiran ideas). In 1964–65 such notions were attached to the supposition that one or two women in each clan were cannibals who could turn themselves into dogs in order to rob newly made graves of their corpses and consume the flesh. ⁵ The propensity was believed to pass from mother to daughter. Historically, there was a greater emphasis on such notions in the northern Melpa area close to the Jimi Valley, and in many ethnohistorical representations of witchcraft the Jimi is cited as its origin place. The Melpa who live in the Jimi Valley are bordered on the north by the Karam, among whom witchcraft ideas are common and are called by the term koyb, the same as the Melpa koimb. Kum koimb is Melpa for “cannibal witchcraft,” thus the concept may be a fusion of ideas of kum, shared with the Wahgi and Simbu peoples to the east of the Melpa area, and koyb coming from the Jimi area to the north. This may help to explain why kum is sometimes spoken of as a category of spirits that live beside watercourses or in bush areas, separate from people and sometimes as the force of witchcraft inside people. The witches of the Jimi Valley are believed to be able to enter people’s bodies through their anuses and to eat their way through their intestines, exiting through their upper orifices and leaving their victims as good as dead. The image here is one of assault witchcraft. These ideas also represent the Melpa’s fears of their very different northern neighbors, the Karam.

Michael O’Hanlon describes kum witchcraft among the Wahgi as intangible malevolent powers that certain individuals of either gender have that allow them to cause illness and death in others. Descriptions of witch attacks may stress the intense gaze of the witch or the harboring of small familiars (e.g., a cat, marsupial rat, or a snake) that a witch unleashes against his or her victim. This familiar is often said to emerge through the anus of the witch and it devours its victim’s inner organs. Accusations of kum witchcraft among the Wahgi also tend to occur after an epidemic outbreak or illness, or deaths in an area (O’Hanlon 1989: 57–58). O’Hanlon has reported for the Waghi a heightened number of kum sorcery accusations at the time of funeral feasts, when dysentery epidemics may spread through the attendees. The Wahgis at these funeral feasts said that the people practicing kum should be sent back to the Jimi Valley, indicating that they share the same ethnohistorical traditions as the Melpa. There are traces here of a precolonial history, but they are hard to follow up. We turn therefore to more recent times.

Colonial changes in New Guinea have always been accompanied by epidemiological changes, the flow of sicknesses. In looking at what Dan Sperber (1985) has called “the epidemiology of representations” in terms of the notion of “political space,” we also pay attention to what may be called the “representation of epidemiologies,” that is, physiological events conceptualized in cultural terms seen as invasions that intrude upon and consume the people’s lives. We also look at “social epidemiology,” the representation of changing social patterns in terms of witchcraft and sorcery attacks (cf. Douglas 1991). The epidemiological history of Melpa ideas of witchcraft seems to have started before colonial and postcolonial change and to have represented the contours of relations with neighboring peoples seen as “others” whose practices invaded the Melpa. Ideas have continued to change and develop over time. We compare accounts from 1978 and 1995.

In 1978 a rumor spread throughout the northern and central Melpa areas that more people were becoming cannibal witches. Instead of just a few women in particular lineages, as previously held, now it was feared that many people of both genders were turning into cannibals because the existing witches were secretly placing pieces of human flesh into the headwaters of streams from which people drew their drinking water. The “grease” or fat from the human meat was tasted and experienced as “sweet,” turning people into cannibals. People therefore were enjoined never to drink water outside of their own clan areas and even within these areas to be very circumspect (Strathern 1982). We see here a prime idea of the spread of danger, through the pollution of water and the need to control consumption in order to avert the new danger flowing through the clan areas. We suggest that this was an expression of the collapse of the “properly” bounded clan area as perceived in the postcolonial imagination. The connections with fears of greed and excessive consumption associated with the advent of cash through the growing and selling of coffee and the relative decline of the exchange ethos in Hagen also seems clear here. The solution suggested was that everyone should control consumption of the medium believed to carry danger—the water supply—a solution that mimics epidemiological hygiene. Water is seen in curative rituals as a healing, purifying substance, yet it was also believed to harbor kum stones. As an element flowing between two categories in a landscape, water could take on either healing or harming capacities. For example, a person, who greedily consumed pork at a feast, who slaked his or her thirst at a stream, gave an opportunity to these little kum stones in the water to jump into that person’s throats. In one image these stones are seen as namb and pilamb (“let me eat, let me experience”). They would scrape at their host’s throat and make him or her insatiable for pork to feed them.

We compare this with the epidemiological rumors of 1995 (Stewart and Strathern 1997). Before this, witches had been reported to eat only the dead: they patrolled around burial sites and fed on corpses. In 1995, however, the idea emerged that there were not enough corpses for the witches to feed on, so they needed an alternative source of food (i.e., living people). Ru Kundil, who gave this account, noted, “We’re afraid because we think the witches may eat all of us living people and finish us off. Kum koimb is the name for these witches” (quoted in ibid.). Ru also described the practices of a traditional healer, Toa, an expert in all kinds of magic spells, who had developed a new technique for removing witchcraft familiars from people.

There are three differences between the 1978 and the 1995 narratives. First, by 1995 the cannibal witches had become more aggressive. Second, Ru declared that these witches were said to be controlled by a queen who lived in the Simbu area east of Hagen, which was seen as the home of dangerous people belonging to a different province (another image of postcolonial political space). And third, the situation was so serious that it now supported the entrepreneurial rituals of a witch-finder (who acted like a pest exterminator by trapping the witches’ animal familiars). Comparing the 1978 and 1995 rumors, we see that in both cases there was a fear of an outbreak of excessive desires for consumption. In 1978 this was controlled by hygienic taboos, but in 1995 these taboos were not adequate: the cannibals were turning into murderers and curative magic was invented to restore them to normality.

In 1997 a further development of ideas took place. We found the Melpa to be experiencing a wave of millenarian notions concerning the “world’s end,” in which witchcraft activity was seen to be on the rise because Satan had instilled kum into his followers. In turn, this was taken as a sign that the world would end shortly (Stewart and Strathern, eds. 1997; Stewart and Strathern 1998a). The domination within the world of greed and desire would consume everything and bring about retribution from God and Jesus, since curative rituals could no longer control the situation.

These descriptions of the spread of witchcraft are reminiscent of detailed accounts of how actual microbial diseases such as typhoid spread through a population. It is thus quite interesting to consider whether the spread of fears of this kind may be at least correlated with changes in disease patterns. The suggestion is theoretically plausible although impossible to demonstrate. Epidemics of dysentery entered the Hagen area in the 1940s but were not attributed to witchcraft or sorcery (the same is true in the 1950s among the Duna). An outbreak of madness in Pangia in about 1960 was said to have followed from a dysentery epidemic that killed up to 25 percent of some village populations, but it was not explained as either witchcraft or sorcery (Strathern 1977). Nevertheless, it is possible that the 1995 fears were at least partly generated out of an increase in deaths from typhoid fever that was in fact occurring at the time, and that the earlier ideas about the pollution of water were echoes of public health lectures about the need to avoid defecation in streams so as not to spread germs, including the typhoid-causing organism. ⁶

With the spread of malaria and the perceptions of witchcraft, we are on firmer ground if we retrace our steps and look again at connections between the northern Melpa and the Karam people as Inge Riebe (1987: 212; see also Riebe 1991) has described them. Riebe argues that Karam people were said to have died of koyb witchcraft from the end of the nineteenth century onward and that the technique was introduced into their area from the north, the Ramu flats. The koyb was thought of as a small snakelike creature kept in the abdomen of the witch, enabling its human host to kill others. Further details were added to this belief through contact with Ramu peoples; koyb witches had the ability to “change into animals, or other humans, become invisible, move at incredible speeds, or be in two places at once, kill without contact, and to sew together and temporarily resuscitate people killed with conventional weapons. Witches were also said to have a greed for human flesh” (Riebe 1987: 214). Witches were thought of as stingy and greedy people and—in an important twist—“were thought only to kill when paid to do so by normal humans” (ibid.).

Koyb witchcraft was given as an initial explanation for sudden deaths. Riebe (1987) suggests that this correlates with the entry of deaths from dysentery and malaria. Settlers from ethnically different northern valleys were more likely to be thought of as witches. They are likely have brought new diseases from their lowland areas into the highlands. Riebe further points out that accusations of witchcraft following such deaths were at first most often made against matrilateral cross-cousins who possibly had competing claims on land resources. This led to avoidance between the parties involved. Witchcraft accusations subsequently became the basis for negotiated payments of compensation between groups and therefore for the careers of aspiring leaders or “big-men” (ibid.: 218). At the psychological level, Riebe points out that proper social behavior depended on an ethic of generous giving, but “the underside of this world was the world of the witch—the world of greed, destructiveness and extortions under threat of witchcraft” (ibid.: 221). She also adds that accusations of witchcraft took the place of revenge homicide, since “all deaths caused by human agency had to be revenged” and a large proportion of deaths of active adults were now attributed to witches. The witch phenomenon was therefore an epidemic resulting from closures and openings caused by the political space of colonialism and indigenous responses to it as well as to new trackways of disease (on the spread of suanggi ideas in Irian Jaya, see Haenen 1997 and Oosterhout n.d. on Inatwatan ideas).

Riebe’s analysis meshes well with the historical situation of the Melpa. The increased passage in colonial times of notions of koimb (koyb) from the Jimi Valley into central Melpa coincided with travel by Melpas on government patrols into the Jimi area in the 1950s and beyond. The later epidemics of witchcraft fear among the Melpa are also correlated with historical perceptions of growing tension over the inequalities between people, marked by capitalist-style consumerism, as we have suggested, further compounded in 1997 by millenarian motions.

Duna Witchcraft and Assault Sorcery

Witchcraft may begin as revenge and end in politics, as we have seen in the Karam. It can also be an idiom for gender politics, as materials from the Duna speakers of Lake Kopiago in the Southern Highlands Province show. We conducted our fieldwork with some one thousand Duna speakers belonging to five distinct parishes in the Aluni Valley. Like the people of Pangia and Hagen, the Duna cultivate gardens and rear pigs, but they have little cash cropping. These people classify assault sorcery in the same category as female witchcraft. Both phenomena are described as tsuwake, assault sorcery being tsuwake tene and witchcraft tsuwake kono. Since “pacification” and missionization in the 1960s, most other forms of sorcery practiced by males as a part of interparish hostilities have been abandoned and are not an object of fear. In this political space cleared of other kinds of mystical violence, however, ideas about both types of tsuwake have flourished. During 1991 and 1994 assault sorcerers were considered to be crossing the Strickland River between the Duna and the Oksapmin areas and menacing the outskirts of Duna settlements in the Aluni Valley; and internal fears of female witches (often with Oksapmin origins in their genealogies) led to suspicions and occasionally accusations. These two kinds of attack by unseen “others,” internal-female and external-male, had again effloresced in people’s imaginations precisely because of postpacification conditions. Cultural bricolage was also at work: One informant who declared that he had trained squads of youths to be tsuwake tene said that instead of the traditional sago-spine dart employed to pierce a victim’s breast, he was able to use syringes from his trade as an aid post health orderly. In 1998 we were told that the assault sorcerers had mostly been shot with guns in retaliation for their raids, but that the remaining few were now improvising methods to lure victims into their power, by persuading women to act as sexual decoys and then closing in for the kill. The sorcerers were always described as being from the Oksapmin area (that is, as prime examples of “the other” to the Duna). ⁷

Gender politics were at work in two ways. First, “foreign” male assault sorcerers were thought to be threatening local women, making them nervous outside of their village areas and requiring men to stand guard over the women as the men formerly did during warfare, so as to take revenge for any attacks. Second, female witches were thought to be internally threatening villagers because of their desire for “meat.” This idea was used to discourage women from meeting to talk together or to complain about pork distribution, so as to reassert male control and surveillance. Both processes may therefore be seen as postpacification responses by males in an attempt to reinforce their authority over females. Another detail is significant. In precolonial times female witches were said to have been sent as agents by male leaders to kill and consume males in enemy areas. But with pacification, it was believed, female witches ceased to be used in this way and instead acquired an independent volition, allegedly turning against victims in their own communities. This is a historical narrative that parallels in muted form the narrative of Mekeo sorcerers given by Trompf (see the previous discussion in this article).

The underlying issue, as with the Melpa, is the perception of appetitive desire. Pork is highly prized and in precolonial times men excluded women from consumption of the cuts of pork that the men themselves consumed in the course of a roster of religious cult performances. With the demise of these cults, the introduction of Christianity, and the loss of communal men’s houses, however, men found it harder to deny pork to women. Correspondingly, we suggest, men began to accuse women more vigorously of witchcraft. In a further twist, by 1994 certain prominent men whose wives were rumored to be witches would specialize in turning up at pig-kills and expecting to be given meat to share with their wives. Their hosts, afraid of the possible witchcraft of these women, would give especially generous portions to the husbands. The gender conflict over consumption was thus resolved by these men in terms of an alliance coalition consisting of big-man and witch-spouse. In 1994, in the village of Hagu, men also rebuilt their communal men’s house, and these itinerant pork-seekers would overnight in this house to share at least a portion of their meat with its regular male residents and establish ties of male sociality and solidarity.

By 1998 the situation had further developed in more complicated ways, which must be understood in terms of the dynamics of historical change since the time just before the first regular administration patrols in the 1950s and the 1960s. Epidemic outbreaks of disease that were spread through the intrusion of European outsiders and their workers from other parts of the country caused widespread dislocation and depletion of groups among the Duna. In the low-lying areas immediately near the large Strickland River, which marks the boundary between the Duna and the Oksapmin peoples, a number of small intermarrying groups fought among themselves, both before and during this time. Many of their members migrated up-valley to the mountainside parishes where they had ties through cognatic kinship or marriage. The composition of these parishes thus became more heterogeneous than before.

Furthermore, these immigrants came from areas that were identified with the origins of witchcraft. Indeed, the conflicts between the Strickland groups were partly, according to the ethnohistorical accounts, about the accusation of witchcraft. In 1991 and 1998 informants in the parishes of Aluni and Hagu indicated that witchcraft had originated from the actions of a male earth-demon who had emerged from a rock in the Strickland area and had proceeded, according to the origin myth, to seize and consume people. The male demon was persuaded by a woman of one of the local groups, the Makalan, to have sex with her instead of killing and eating her; she then became the host to his cannibalistic propensities that were passed down to her children and descendants (the emphasis here is placed on the transmission through females). The women carrying witchcraft powers married into local groups and their descendants carried these powers with them in the precolonial shift from the environs of the Strickland grasslands to the forested mountain parish areas. Among the Duna it is thus a male spirit who brings the knowledge of witchcraft to women; but it is a female spirit, the Payame Ima, who brings the knowledge to men of witch divination (Stewart and Strathern 1998b). The Payame Ima was not the spirit who brought witchcraft to humans. Rather, she is said to enter into and possess the men to whom she grants ritual power to divine for witches.

Duna parishes are local collocations of kin, affines, and associates, conceptually organized around a body of coresident males descended agnatically from a parish founder. This core is known as the anoagaro, the paternal line, distinguished primarily from those who are imagaro, belonging to the parish through maternal ties. The effect of parish incursions by immigrants was to put the anoagaro in the minority in their own parishes and therefore to make the exercise of their leadership more difficult. For example, they are supposed to control the disposition of land, and this task would be made harder with the absorption of newcomers, including those thought to be witches. At the same time a good proportion of these anoagaro leaders were polygynists, with at least one wife thought to be a witch.

In 1991 and 1994 fears of the actions of these witches appeared to be on the increase, but the local Christian churches forbade the use of traditional methods of divination to seek out and punish witches thought to have killed and eaten witches. In 1996 this prohibition was broken after the dramatic deaths by fire of two young men from the Aluni parish who had taken part in a hunting expedition to the Strickland. They had lit fires in the grasslands (to burn out wild pigs), which then consumed them. A diviner was hired and pinpointed four local women as the responsible witches, who were said to have confessed to their acts. The women were driven out of the community and their kin paid compensation to the relatives of the two dead men. In 1998 this kind of sequence was repeated in Hagu, with two twists. The deaths that triggered accusations were a result of an epidemic (of malaria, typhoid, and pneumonia) following an unusual period of drought and food shortages. The son of a female witch who had “confessed” to his mother’s actions killed his mother, causing the original accusers to pay compensation for the death. Those who had died were two small children of the only anoagaro leader in Hagu, one of whom was his youngest child and only son, the sole successor to the anoagaro line. The witches were from a family with whom the leader had disputed over land use (see Stewart and Strathern 1998b for fuller ethnographic details).

These dramatic and troubling events thus brought to the fore the results of some fifty years of historical change: alterations in village composition and the dynamics of village leadership as well as the prior alterations in the gendered consumption of valued foods and the latest ecological stresses of famine and sickness. Furthermore, another dimension was revealed: accusations about witchcraft can be a part of arguments among anoagaro leaders about legitimate control within the parish. These leaders may accuse others of subverting their control on their line of reproductive descent. But the leaders in different parishes also harbor resentments against one another because of the idea that they have witches as wives. While a leader may use this attribution to coerce others into giving him pork, he is also vulnerable to accusations by rival leaders and others that his wife has killed and eaten someone within his parish or a different parish. Accusations of witchcraft, like those of sorcery, can thus become a mode of politics, entwining issues of gender as well as interparish and intraparish conflict.

Cannibalism, Consumption, and Change

We have seen in the Duna narrative that in the origins of witchcraft a male earth-demon (tama) emerges from a hole and seizes and eats people. This direct form of cannibalistic consumption is then mediated by the earth-demon’s sexual relations with a human woman, from which witches emerge, born from their union. The same myth explains that in those early times the people at the Strickland River ate people instead of pigs. When a pig died, it was placed on a burial platform and mourned; a human might be taken as a captive to be killed and eaten as a funeral sacrifice for the pig’s death. It is told that a wandering hero from a different area came upon a woman about to be sacrificed in this way, and he saved her by cooking the pig instead, offering its flesh with salt (which he had brought as a gift from his own area) sprinkled on it to the mourners; the mourners were thereby persuaded to abandon cannibalism and eat pork instead.

The image of cannibalism here is cognate with witchcraft, since it belongs to the same myth as the story of how witchcraft first developed. Both witchcraft and cannibalism therefore are forms of inappropriate consumption, and progress is marked by their abandonment. The problem for the Duna is that while straightforward cannibalism is seen as having been transcended by the gift of salt in the myth, witchcraft is still thought to be passed on within the community (Stewart and Strathern 1998b).

Narratives of an implicit shift away from cannibalism and its ritual transcendence in other forms of behavior belong to a wider class of origin stories in which “civilization” is presented as a mode of evolution in sacrificial practices. Michael W. Young’s (1983) discussion of victimage in Kalauna fits in here. Victimage in Kalauna (on Goodenough Island, in Milne Bay Province, Papua New Guinea) is institutionalized in two principal modes: a projective system of vengeance, homicide, and sorcery, and an introjective system of self-castigation. “Although these types of victimage may sometimes appear in pure form as vicarious sacrifice and self-sacrifice respectively, they are often found in combination” (ibid.: 29). The legend of the cannibal warrior, Malaveyoyo, culminates in his sacrificial death, which serves as a marker of the beginning of the colonial period when abutu, competitive food exchange, altered the previous pattern of vengeance and violence, serving as a surrogate for killing and eating one’s enemies. Abutu retained the idiom of oral aggression but displaced the object of consumption from the flesh of one’s enemy to food such as pigs that one had reared (ibid.: 92–109).

Exactly the same logical structure is shown in the conventional Hagen statement by relatives of a victim to the killers: “You killed our man but you did not eat him or taste anything sweet, so we give you back pork as his ‘bone,’ and you can give us back pigs for this ‘head’ later.” The figurative phraseology here does not rely on a trope of past versus present but on a notional cannibalism juxtaposed against a real noncannibalism. The message, however, is the same as in Kalauna: The exchange of pork transmutes oral aggression into sociality; put otherwise, it is the conversion of “feasting on my enemy” into “feasting with my enemy.” Images of cannibalistic witchcraft show that very transition arrested or denied, or nowadays a “regression” to a wild state that threatens contemporary life with its return when the life of order through exchange no longer can control the disordered spatial forces of postcolonial turbulence.

Ideas of the spread of assault sorcery also reflect these alterations in spatial relations, since assault sorcerers, like witches among the Melpa, are thought in Pangia and the Duna area to be spreading more widely than before, reflecting the facts of greater movements of people and continuing distrust among them. The assault sorcerer destroys the community by marauding it from the outside, while the witch destroys it by eating victims within the community. Both images reflect the complex changes in the meanings of “the outside” and “the inside” in an altered world of spatial relations.

Conclusion

Our arguments in this article have been presented in our account of historical changes in three different areas of the Highlands of Papua New Guinea, the Pangia, Hagen (Melpa), and Duna areas. The materials from these areas differ somewhat in their historical scope and focus. Pangia does not appear to have notions corresponding exactly to those of cannibalistic witchcraft such as are found in Hagen and among the Duna; therefore, this article discusses only the case of assault sorcery. Hagen lacks the exact notion of assault sorcery, but kum koimb witchcraft partly includes images comparable to those found in Pangia. Among the Duna there is both assault sorcery and witchcraft ideas.

This disparity of ideas, combined with areas of overlap, does not affect our main argument. We show that the wider field of assault sorcery and witchcraft can be examined to reveal alterations in scale and intensity of the kind we explore. These alterations involve some shifts and innovations in cultural tropes or images—for example, the Hagen idea that witches hid pieces of human flesh in streams or that witches in 1995 were attacking living people, not just corpses; or the Duna idea of the substitution of syringes for sago-spines and the use of women as decoys to attract male victims. These alterations in images can themselves be traced to various changes in the people’s way of life. Our main focus, however, is on changes in the contexts of operation of these “mystical” ideas and the ways in which they go with larger patterns of change. Several factors combined to produce these alterations: the movements of people, the mixing of linguistically or socially different groups, the effects of disease, alterations in group composition and the condition of leadership, changes in patterns of inequality arising from patterns of cash cropping and the consumption of goods purchased with money, and shifts in the balances of power between the genders and between local leaders. In this mixture of factors we stress both the “epidemiology of representations” in terms of images of violence, and the “representation of epidemiologies” of sicknesses such as malaria and typhoid. Finally, we argue that colonial “pacification” and postcolonial instabilities combined to produce perceptions of political space in which notions of sorcery and witchcraft, far from disappearing, mutated and spread.