The power of local ties: Mechanisms of mass participation in the Rwandan genocide

Lee Ann Fujii Department of Political Science George Washington University lafujii@gwu.edu Paper1 presented at the Annual Meeting of the Midwest Political Science Association, Chicago, IL, 20-23 April 2006

By any measure, the Rwandan genocide was a catastrophic event. In less than 100 days, killers killed over half a million Tutsi and tens of thousands of moderate Hutu. The architects of the genocide were a powerful group of Hutu extremists intent on defying the power sharing terms of a recently signed peace agreement between the current regime and the Rwandan Patriotic Front (RPF), a guerilla army comprised mostly of Tutsi refugees, which had invaded Rwanda in October 1990. Through genocide, the extremists sought to maintain their monopoly on power. The killings began on 6 April 1994, shortly after unknown assailants2 shot down the plane carrying President Habyarimana as it tried to land at the airport in Kigali. Within hours, specially trained militia, soldiers, and Presidential Guard began going door to door with lists of Hutu and Tutsi targets, whom they slaughtered with gruesome efficiency. In the rest of the country, however, the genocide followed a different path. Political entrepreneurs in rural communities took the opportunity to secure their own power by enlisting local residents into genocide. Despite long-standing ties with their victims, tens of thousands3 of ordinary peasants hunted and killed their Tutsi neighbors with machetes, hoes, and clubs. How does such participation become possible? What are the mechanisms for recruiting ordinary people into mass murder?

Foils and frameworks

The usual approach to explaining genocide and ethnic violence is to point to ethnicity as the key driver, particularly at the level of non-elites. Two of the most prominent ethnicity-based explanations focus on ethnic hatreds and ethnic fears. The ethnic hatred thesis, trumpeted by journalists and scholars alike, posits that mass
This paper is part of a book-length project entitled Killing neighbors: Social dimensions of genocide in Rwanda. The data were collected in 2004 during nine months of fieldwork in two rural communities and central prisons in Rwanda (Ruhengeri in the north and Gitarama in the center-south). The fieldwork involved multiple, intensive interviews with a purposive sample of local residents who participated and/or lived through the genocide. The sample included confessed killers, former prisoners, current prisoners who maintain their innocence, rescuers, resisters, witnesses, and survivors. To protect the identities of those interviewed, I use pseudonyms for people and places and do not include dates for the interviews. 2 The identity of the assailants is still uncertain. Prunier (1997) argues convincingly that it was the extremists in Habyarimanas own inner circle who ordered the assassination. More recent reports, however, one by a former member of President Kagames inner circle, declare that it was Kagame himself who ordered the plane to be shot down (see, e.g., ___). 3 Straus (2000, 94-95) estimates between 175,000 and 210,000 total perpetrators, of which 90% would have been non-hardcore civilian perpetrators.
1

violence is the product of extant ethnic hatreds, which erupt into violence when the opportunity presents itself. In this view, hatred is an integral part of ethnic group identity that never fades, transforms or alters; it simply simmers until someone or something pushes the lid off the boiling pot (Kaplan 1994; Kaufman 2001; Posen 1993). The ethnic fear thesis focuses not on cultural constants, but on elite ambitions and actions. Elites foment fears of the ethnic other in pursuit of their own political goals and interests. Ethnic fears are thus a creation of elites, not an extant feature of ethnic group identity. Ethnic groups nonetheless heed or support elite appeals to commit violence because it is rational4 to do so (Posen 1993; de Figueiredo and Weingast 1999; Hardin 1995). While both approaches look to ethnicity for explanatory leverage, neither approach explains patterns and characteristics of violence that occur outside capital cities. First, neither approach makes a distinction between those who actually participate in such campaigns and those who do not. Yet, this distinction is critical, for while it is true that tens of thousands of ordinary peasants took part in the genocide in Rwanda, it is also true that hundreds of thousands did not, despite being subject topresumablythe same ethnic hatreds or fears. Why? Second, neither approach can account for the social intimacy of violencethe fact that ordinary people do not simply participate in any violence, but specifically in violence against people they know (which is why leaders enlist locals in the first placebecause locals know who the local Jewish, Tutsi, or Muslim families are and who their sympathizers are). Indeed, the literature on social capital would expect long-standing, cross-ethnic, reciprocal ties (such as those that typified community relations throughout Rwanda) to constrain neighbors from committing violence against. Why did these ties not operate as expected? This paper addresses both these puzzles. It begins by disaggregating the masses to identify the key groups of actors who did and did not participate in the violence. It then proceeds to analyze the social context in which these actors joined in the violence. The papers findings are based on data collected in two rural research sites in Rwanda: Kimanzi secteur in the northern province of Ruhengeri and Ngali secteur in the central province of Gitarama. The paper argues that far from constraining violence, social ties facilitated local participation. Vertical ties served as mechanisms of targeting and recruitment, while horizontal ties brought together the lowest-level participantsa group I call Joinersenabling these actors to respond to the genocide as a group. ***** The starting point for this investigation is the work of sociologist Mark Granovetter (1985) and his concept of social embeddedness. Granovetters argument is simple but powerful. He argues that most behavior is closely embedded in networks of interpersonal relations and that such an argument avoids the extremes of under- and oversocialized views of human action (Granovetter 1985, 504). Undersocialized approaches treat actors as pure agents, unfettered by any social ties; while
In Posens application of the security dilemma to ethnic violence, it is rational for any ethnic group to try to increase its security when the state has collapsed or can no longer guarantee its citizens security. In de Figueiredo and Weingasts argument, it is rational to accept what elites say about the demonized group since the risk of not accepting their words is possible annihilation (if elite prognostications turn out to be true). Finally, it is rational in Hardins framework for people to identify with their ethnic group because ethnic groups provide coordination power and coordination power enables groups to attain their goals, which are also the individuals members goals.
4

-2-

oversocialized treat social ties as structures that determine all. For Granovetter, the notion of social embeddedness strikes a middle ground. As Granovetter argues, social relations have an independent effect on actors, an effect that is neither merely frictional nor overly deterministic. For Granovetter, systems of social relations can and do act as sites of trustbuilding (as the literature on social capital teaches; e.g., (Varshney 2002, 1997; Putnam 1993; Coleman 1988), but this does not preclude the possibility that distrust and malfeasance can arise at the same time. Granovetter points out two reasons for this seeming paradox. The first is that by trusting someone, people become more, not less, vulnerable to betrayal. The second is that malfeasanceparticularly of the more ambitious varietygenerally requires teams or groups of bad-doers; large-scale bad deeds simply cannot be accomplished alone. Thus, the extent of disorder resulting from force and fraud depends on how social relations are structured or organized. While an elaborate heist might require a small group of thieves, for example, a war requires much larger numbers of people organized into social units called armies. As Granovetter explains: More extended and large-scale disorder results from coalitions of combatants, impossible without prior relations (Granovetter 1985, 492, emphasis added). Extending Granovetters concept of social embeddedness to an analysis of the violence during the Rwandan genocide is useful from several angles. First, such an approach invites us to disaggregate the masses and to differentiate between groups of key actors. Second, it leads us to explore the structure and content of relations that tied these groups together in concretethat is, consequentialways. Third and finally, it provides a way of linking the configuration of relations to behaviour not in a mechanical or deterministic way, but in a way that takes social relations seriously as mechanisms that produce specific types of behaviour in specific social and political contexts.

Parsing the masses

In genocide scholarship, the standard categories for sorting actors are perpetrator, victim, bystander, and rescuer. This taxonomy generally assumes that membership is constant and exclusivethat actors will fit one and only one box at a time and that the box they do fit will not change. Data from my field research, however, challenge these assumptions in several ways. The data show that the dynamics of the genocide were complex and at times, paradoxical. Killers did not just kill. Many tried to save and hide Tutsi friends and neighbors even as they remained active in the genocide of other Tutsi. Not all Tutsi were killed; some were spared; some even joined the killing groups. Not all victims were Tutsi, some were Hutu targeted despite their ethnicity. The standard categories of perpetrator and victim lead us away from these grey zones of activity by occluding the range of variation within categories and by removing the possibility that actors might move between categories or occupy more than one category at a time. These standard categories, in short, lead us away from the very complexity that typified the patterns of participation in the two research sites. Given this level of complexity, who were the key actors at the mass level? I make two cuts at parsing the masses. The first cut distinguishes between those with power and those without. Power relations are always important, but no more so than during periods of heightened fear and insecurity, such as the genocide. Those with power had different options. They were in a position to steer events and shape perceptions. They were able to coerce, cajole, entice, or persuade others to follow their lead. Those without power faced a different set of constraints and opportunities than power-holders. This did not make the powerless mere followers of the powerful, however. Neither power nor ethnicity, as I will argue, determined a persons fate, even during genocide.

-3-

Among those with power are two groups: leaders and collaborators. Leaders were local elites who derived their power from connections to higher ups in the dominant political party or the government hierarchy. They would have cultivagted these connections before the genocide began and then capitalized on them when the opportunity arose. Indeed, leaders generally had a history of positioning themselves closely to those more powerful than they, yet these attachments were not characterized by ideological commitment, but by strategic calculation. Leaders seek power. Connections to the more powerful are means, not ends. The genocide, too, was a means for local leaders to establish or consolidate their own power. Thus, after the assassination of the president, leaders were the first to begin organizing local residents for violence. As part of this effort, leaders enlisted lieutenants who helped to realize leaders orders and plans. Finally, leaders used their position to demobilize resistance to their program. Indeed, demobilizing resistance was just as important as mobilizing support.5 One form of support came from the second group of powerholders collaborators. Collaborators derive power in the momentfrom aligning themselves with leaders and their programs, either directly or indirectly. Like leaders, collaborators seek to profit from the moment, and thus, are often motivated by personal, not ideological, interests. One collaborator, for example, in the northern research site of Kimanzi, used his connections to the bourgmestre6 to have a local Hutu man imprisoned twice during the civil war, accusing the man of being an accomplice of the RPF. Following the end of the civil war, this same collaborator then worked with the RPF authorities to have the same man imprisoned a third time as a gnocidaire. Collaborators are not lackeys of leadersthat is, they do not simply do leaders bidding. Rather, they use leaders and leaders programs to pursue their own private or personal interests. Among those without power are two groups. The first group I call Joiners. Joiners were the lowest level participants in the genocide. It is they who had the most to lose and the least to gain by participating in the violence; thus, it is they who represent the core puzzle of the genocide. What characterizes Joiners is not a single set of motives, but their direct participation in the violence. In my two research sites, Joiners were responsible for committing the majority of physical violence against Tutsi and other targeted victims. In contrast to Joiners, leaders, and collaborators are those who did not participate in the genocide in any way. This group I simply label non-Joiners for they encompass a wide range of different actors, including those who were the main targets of violence (survivors); those who helped to save or rescue Tutsi (rescuers); those who evaded participation by hiding, paying off a leader, or by paying someone else to go in their place (evaders); those who remained neutral, neither trying to help nor hurt anyone (bystanders); and people who refused or resisted pressures to participate in the violence in any way (resisters). I combine this group of heterogeneous actors not because their behaviour is explained by a common set of factors, but because the focus of this paper is Joiners. Thus, it is Joiners actions that I foreground against the background of all those actors who did not join in the violence.

The demobilization argument comes from Gagnons (2004) excellent analysis of the wars in the Balkans. 6 Until quite recently, Rwandas administrative structure was comprised of eleven prfectures (each headed by a prfet). Each prfecture, in turn, was comprised of communes, each led by a bourgmestre. In this structure, bourgmestres wielded a great deal of power at the local level.

-4-

Ties that bind

What linked actors of various levels of power and influence, from leaders to Joiners, was a dense set of overlapping ties. These ties did not reduce to objective categories of mother, brother, or conseiller. Rather, they were the sum of shared understandings and activities, daily interactions, and regular face-to-face contact. They were the product of talking, visiting, sharing beers, attending meetings, and participating in umuganda, the mandatory communal work requirement. It is both the fact of these relations as well as their content that shaped how people responded to the genocide. In both communities, social ties crossed ethnic lines. Among the eight survivors in my sample from Ngali,7 for example, half were married to Hutu spouses at the time of the genocide. Many killers from Ngali also had Tutsi family members through marriage. Even the local leader of the genocide in Ngali (discussed in detail in the next section) turned out to have a formerly Tutsi wife; that is, his wifes parents and paternal grandparents had changed their ethnicity from Tutsi to Hutu in 1959 to avoid persecution as Tutsi. In addition to ties through marriage, Hutu and Tutsi shared longstanding, close friendships in both communities. Even in the northern research site of Kimanzi, where the number of Tutsi was very small (approximately 40 Tutsi for the whole secteur as opposed to 200-300 Tutsi for Ngali), close friendships between Hutu and Tutsi were not uncommon and were even expressed through the symbolic ritual of a blood pact. Despite the ties that connected people across ethnic lines, the same set of ties made people vulnerable to recruitment into the genocide or victims of the genocide. In both communities, the key ties were those that emanated to and from the central leader or leaders of the community. Those situated closest to these centers of power were the most likely to be recruited into genocide. Let us look in more detail at the specific configuration of ties in each community. Single center of power in Ngali (center-south) In Ngali, there was a single center of power, a man I call Jude. Jude was the undisputed leader of the Interahamwe,8 the term people use to refer to the groups of local killers. Though Jude had died in 1997,9 there was no shortage of information that residents of Ngali were able to provide. By all accounts, Jude wielded a monopoly on the means of coercion. Ngali residents variously describe Jude as the head of the Interahamwe, president of the Interahamwe, the head of Pawa,10 the head of the killers, and our leader. In his quest to consolidate power, Jude enlisted the help of a small group of lieutenants who worked with Jude willingly and enthusiastically. The seminal event in Ngali was the plane crash that killed President Habyarimana on 6 April 1994. Before that moment, daily life had remained normal in Ngali, despite the civil war that had begun in the north of the country with the RPF invasion on 1 October 1990. Judes actions in the aftermath of the assassination were swift. As Paul, 11 Judes nephew, explained:
The total number of people I spoke with in Ngali was 42, of which 8 were Tutsi survivors. The sample was purposive, not random, and thus not suitable for statistical analysis. 8 I use this term following local usage. It is worth noting, however, that the term originally referred to the militia that the government recruited and trained in Kigali. 9 According to Judes wife, Jude died shortly after he and his family returned from exile in the Congo in 1997. Jude was taken directly to the local jail (cachot communal) and died 3 days later. His wife never specified the cause of death. 10 Pawa is a derivation of the English word power. In 1993-1994, Pawa referred to the extremist wings of the various political parties that had emerged at the time. 11 At the time of my fieldwork, Paul was still in prison. He was among those in my sample of
7

-5-

He named himself [conseiller or head of the secteur of Ngali] on April 15, 1994. He was coming to the JOKOMA center.12 He told me that I was no longer conseiller because the Gitarama authorities had just appointed him, and that he was going to change all the responsables. The following day, as he was president of the MDR13 party, he called the AbaJDR (the youth wing of the MDR). For each cellule, he named the responsables who belonged to his MDR party [Paul, #2/414]. Judes political party connections were crucial to his rapid rise to power. His backing by more powerful patrons, chief among them the bourgmestre for the commune where Ngali was located, guaranteed his usurpation of power. Judes rise may also have had some support from below as well. Judes wife provides a more populist view of how Jude assumed a leadership role after the presidents plane crash. My husband was the responsable under the government of Habyarimana. When the war of 1994 started, everyone came to look for my husband to ask him his advice. Because of all his advice, my husband became the leader or conseiller of the secteur of Ngali. When people started to hunt down the Tutsi, they would pillage their belongings and they would bring all these pillaged goods to our house. After noticing that these Hutu were bringing all these Tutsi belongings to our house, my husband would ask them to take their looted things to their own homes. When they started to kill the people, I asked my husband: Why are these people killing these Tutsi? What did they do? My husband told me that it was the order coming from the high authorities. I was scared and I became ill. I put myself to bed until we fled toward the Congo [Thrse, #7/7]. This summary of events emphasizes the leadership role that Jude already held in the community prior to the assassination of the president, and his ability to capitalize on that previous position of authority. Indeed, as a skilled entrepreneur, it is quite likely that Jude cultivated both sets of connectionsthose to officials higher up in the MDR, the locally dominant political party, as well as to the people in his cellule and, to a lesser extent, those living in other cellules. According to numerous testimony, Jude was adept at using force and terror to ensure compliance to his wishes. His wifes testimony, however, suggests that the relationship between Jude and to some residents in the community
prisoners who maintained their innocence and thus, had elected not to confess. 12 Though I never found out what the name meant, the JOKOMA center was a central area on the main road that runs through Ngali. In 2004, it was marked by no more than a small cabaret (a local bar), with houses running along each side of the road. 13 MDR or Mouvement dmocratique rpublicain was the reincarnation of the MDR-Parmehutu party that was dominant under Rwandas first post-independence president, Grgoire Kayibanda. Habyarimana subsequently banned the party in 1975 shortly after he took power in a coup dtat in 1973. Politics in Rwanda has always been regional in nature, not ethnic. Thus, the reason why Habyarimana banned the party was to unseat powerholders who came from Gitarama, Kayibandas prfecture, and to replace them with people from his own northern prfecture of Gisenyi. For more information on the original MDR-Parmehutu party, see Lemarchand (1970) and Newbury (1988). For background on the founding of the MDR in 1991, see Bertrand (2000), especially Chapter 3. 14 The nomenclature #X/Y denotes from which interview, out of the total number of interviews with the particular person, the quoted passage came. I have removed dates to ensure anonymity.

-6-

may not have been based completely on coercion or force. What is clear is Judes power position vis--vis his nephew, Paul, whom he ousted as conseiller. People consistently describe Jude as all powerful and Paul, as powerless. As one confessed killer remarked, At that time, Paul was incapable of making decisions. We saw that Jude had replaced Paul. It was Jude who was powerful [Olivier, #2/5]. An older Hutu resident of Ngali went further: He was a leader. He would say all the time that he was working with the authorities of the province of Gitarama. He would do whatever he wanted, no matter where, no matter when, at no matter who [Bernard, #2/6]. I pressed Judes nephew, Paul, for more information on their past relationship to glean a broader context for these events. Was Judes move to name himself conseiller strictly political, or was there something more personal behind this move? Pauls response hints at both factors. Q. How was your relationship with Jude before 1990? Jude was first of all my maternal uncle. He was also a responsable of one of the cellules of my secteur. Q. Did you know Jude from childhood? I knew him very well since I was a child. Q. While you were growing up, did you see Jude often? I was very young compared to him, but I saw him often because he got married when I was still a teen-ager. I would share a beer with him at the local bar and at the house. He would also come over to my house often. Q. And his wife? Did you know her well, too? I knew her well because she would come say hello to my wife each time. Q. Did Judes wife participate in his activities, too? No, she was traumatized seeing what her husband had pillaged and filling the house [with] cows, goods of all kinds. Q. After having named himself conseiller, did Jude threaten you? There was jealousy because he thought he should have been conseiller. After he was named conseiller, I would not give him the registry that helped me give Tutsi [a paper] so that they could go anywhere. Q. When the killings started, did Jude target specific people? Yes, all the Tutsi and their supporters (Hutu) and everyone who looked Tutsi. Q. Among all Tutsi of Ngali, did he target specific people? I am saying all Tutsi without distinction [Paul, #2/4]. While Paul seemed to have been unafraid of Jude, even after Jude took over as conseiller, it is not clear what Pauls relationship was to Jude during the course of the genocide. On the one hand, Paul remained a great threat to Jude because he defied Judes orders to kill Tutsi, and instead, helped many Tutsi escape by issuing them papers stamped with the official secteur seal (stamped documents carrying great weight in Rwanda). The document stated that the document-holder was Hutu and had lost his or

-7-

her identity card. With this paper, local Tutsi were able to flee the area. For these actions, Jude threatened Paul on numerous occasions, calling him an icyitso (accomplice) and wrote up Pauls activities in his weekly reports to the communal authorities. There is no indication that Jude ever roughed up Paul, however, or ever threatened to kill him. Neither did he ever imprison Paul. In fact, Judes and Pauls relationship appears to be ambiguous, wavering between rivals and colleagues. One survivor, for example, explained how her husband approached Jude to protect her from a relative who had threatened to kill her: My husband went to Jude. He gave him some beer and Jude asked Paul to give me a paper that said that I was Hutu [114, #6/7]. This womans statement suggests that the relationship between Jude and Paul may have entailed both collaboration and resistance. What this survivors statement (corroborated by others testimony) also indicates is that Jude was no ideologue. In addition to sparing this womans life, Jude also allowed several young Tutsi men to join the Interahamwe after the men paid Jude various amounts of money. At the same time, Jude was not reluctant to use force or intimidation to dampen resistance or cajole participation. The use of force was a clear expression of Judes power. Discretionary use of that forcesuch as allowing certain Tutsi to live underscored the absolute nature of his power in Ngali. ***** Jude was clearly the center of power in Ngali and ones social proximity to Jude made one more likely to be recruited into the Interahamwe. For Jude, genocide started at home. Jude conscripted his whole family into genocide, including cousins and extended family, and threatened to kill those who did not go along with his plans. Family was a source of ready recruits for others as well. One of Judes chief lieutenants, a man whom one neighbor referred to as Judes advisor, targeted family members as both recruits and victims [Bernard, #2/6]. This man killed the Tutsi children of his sister and then tried to kill the sister as well. He also forced his nephew to join the Interahamwe. A survivor, who had been friends with the son, explained the sons forcible recruitment: Q. When you say that your friend became an Interahamwe because of his uncle, what do you mean? that his uncle forced him to join them? He forced the child to be in the group of Interahamwe. Even my cow was killed by the [uncles] group. My friends father did nothing during the period of the genocide and there is another paternal uncle who did not participate in the genocide. Q. How did [the uncle] force your friend to join the Interahamwe? [The uncle] was the head of Pawa. He was powerful and he told [my friend] to join the Interahamwe and if he did not join them, they were going to hurt him. Q. If your friend refused to join the Interahamwe, would he [the uncle] have killed him? He would have killed him [201, #7/8].

-8-

As this mans testimony underscores, family ties did not insulate people from the danger of genocide; it made them more vulnerable as ready recruits or ready targets. The fear that genocide generated at the local thus tended to follow familial, not ethnic, lines. In addition to direct or forcible recruitment, there were two other ways of targeting family members. These methods were more indirect in that they did not involve killing the target oneself but rather, turning the person into a legitimate target for others to kill. The first method was to denounce a person for hiding or helping Tutsi. In my sample, denouncers went after immediate family members. In one case, a son denounced his father for hiding two children of a close Tutsi friend (the Tutsi friend had died years before the genocide). In another case, a man denounced his brothers wife and wifes family who had sought shelter in the brothers house. In a third case, a man threatened repeatedly to denounce his grand-mother for hiding Tutsi in her house and then finally did turn them in. Because denouncing was a deadly practice, it was a convenient way to get rid of unwanted family members as was the case in all three examples above. The son who denounced his father did so to claim his inheritance. The brother who denounced his brothers Tutsi wife and family wanted to take over his brothers house and belongings. Finally, the grand-son wanted his grand-mothers rescue activities exposed so he could rape the Tutsi girls she was hiding. These denouncers were clearly motivated by personal, not ethnic, motives. A second method of targeting was to accuse a person of being an icyitso (ibyitso, pl.) or RPF accomplice. Like the practice of denouncing, the practice of accusing was also motivated by personal, not ethnic, reasons. One prisoner, for example, already threatened by Jude, faced additional threats from his cousin. As this prisoner explained: Olivier [the cousin] also wanted me dead because of a disagreement between him and me even though my father and his father are brothers. Family conflicts made family ties dangerous and threatening during genocide, not protective and insulating. There were, to be sure, family members who did not participate, despite their proximity to leaders or collaborators. In one notable example, Judes wife, Thrse, did not participate in or even profit from any of her husbands activities despite being in a privileged position to do so. Such reluctance stands in stark contrast to accounts that stress the active role that wives played in pillaging clothes and other belongings off the bodies of their husbands victims (see Hatzfeld 2000; 2003). Thrses proximity to Jude, however, did have an effect on Thrse. Being so closely situated to the center of power made it impossible for Thrse not to know what was happening, particularly as the looted belongings of Tutsi victims piled up in her living room day after day. Thus, despite her attempt to retreat inward as so many others did during the genocide to get by (Wagner 1998), Thrse could not help but know what was happening, as her silence on the subject during our many interviews seemed to indicate. For those less closely tied to Jude, by contrast, not knowing remained a viable option and an option that many Hutu took. In Ngali, in sum, family ties were the most binding of all, for they provided means, motive, and opportunity for family members to target other family members. Genocide, in short, was not just a local affair in Ngali, it was also a family affair. Multiple centers in Kimanzi (north) In Kimanzi, located in the northern province of Ruhengeri, there was no single, center of power but multiple centers. This was due, in part, to the active role that the military, various authorities, and other state agents (e.g., communal police) played in the genocide in Kimanzi. As one confessed killer explained: It was the authorities (the commune, conseiller, responsable) who were telling the entire population to be careful because there were ibyitso. The military, too, especially the military, were really -9-

encouraging the population, saying they shouldnt remain sitting down [Edouard, #2/5]. The conspicuous presence of military in Kimanzi was due to RPF positioned in the mountains ringing Kimanzi; and the rebel armys ability to attack from these positions. The deployment of Rwandan military in this region also provided soldiers with opportunities to kill Tutsi, by enlisting the help of local people. One confessed killer, for example, related the story of a soldier who brought truckloads of Hutu and Tutsi to a remote location in Kimanzi, telling the population that his captives were RPF accomplices. He then ordered his victims to dig a hole and place themselves in it. He then ordered the population to stone the victims to death and finish them off by burying them in the hole. Despite the multiplicity of power centers, the patterns of local involvement in violence in Kimanzi were similar to those in Ngali. Local officials who had connections to the bourgmestre, for example, were able to wield an inordinate amount of power. These officials used this power to intimidate, harass, and threaten local Tutsi and certain Hutu on an ongoing basis. One Hutu man, for example, was imprisoned on three different occasions. The man who had him imprisoned each time was the former conseiller of Kimanzi who had lost his post in 1989 because of accounting irregularities. Even after losing his position, the man maintained his ties to the bourgmestre. After the launch of the civil war in October 1990, he used this connection to imprison many Hutu and Tutsi. Patterns of recruitment in Kimanzi, however, did not start at home as they did in Ngali, but they did start with those within geographic proximity. Leaders of killings often enlisted nearby responsables and the heads of ten households (nyambakumi) to help with the disposal of bodies. Other leaders ordered the population to witness the killings of Tutsi or to take part in the killings more directly. In brief, ties and proximity also mattered in Kimanzi but in different ways than in Ngali. Family ties were less important, but geographic proximity (that is, ties through residency) were still crucial to the recruitment of local residents.

Ties that join

Ones proximity to the center of power was not the only set of ties that mattered, however. Ties between Joiners and between Joiners and non-Joiners were also important in shaping Joiners behaviour during the genocide because these horizontal ties shaped how Joiners understood their situation and the range of options they believed they had in a given moment. Before we explore these ties in more detail, let us draw a more detailed picture of who these Joiners were. Were they simply unemployed or disenchanted youth as some observers would have us believe (Kaplan 199_)? Or were they professional thugs and criminals who had no compunction about killing for profit or sport (Mueller 2000)? In Ngali and Kimanzi, Joiners fit none of these categories. Joiners in both research sites were, in every way, ordinary men of their communities.15 They were all married men with children. Their average age was 32. Most stated their occupation as cultivator or farmer.16 Most were born in the same community in which they lived at the time of the genocide. In Ngali, for example, 9 out of 10 Joiners stated they were born in Ngali secteur. Nearly all mentioned having Tutsi neighbors or friends. Many (albeit
This is also true of participants of riots, both historically and during contemporary times. See, e.g., Tambiah (1996, 216-17) on ethnic riots in India, Pakistan, and Sri Lanka. 16 Among the exceptions were a man who gave his profession as mason and another who gave his as chauffeur and mechanic. Note that it is common for many Rwandans to earn off-farm income through other means. See, e.g., Jefremovas (2002) on how men and women make extra money working in brickyards.
15

- 10 -

mainly in Ngali) had Tutsi parents or family members. None held positions of power with the state above the level of responsable. None had had any military or paramilitary training prior to the genocide.17 (Those I identify as Joiners were prisoners who had elected to confess to participating in the genocide in exchange for a reduced sentence. All but 2 of the 16 confessed prisoners I interviewed in the central prisons in Gitarama and Ruhengeri [out of a total of 28 prisoners in my sample] were still awaiting processing of their dossiers and so had yet to benefit from their confessions. The other 2 had not confessed but had already received sentences of life imprisonment and the death penalty, respectively. All the prisoners had been in prison at least 8 years, some close to 10, by the time of my interviews.) What these demographic data tell us is not only that Joiners were ordinary in every way, but that they also had prior relationships to one another (of varying levels of closeness) before the genocide, as Granovetter would expect. What the following profiles help to reveal are the types of ties that existed between Joiners and between Joiners and non-Joiners. Paul (non-Joiner, Ngali) I first met Paul, Flix and Eugne during a brief group interview. The three were an unlikely trio, differing quite a bit in age, education, and personality type, but they presented themselves as a group of close friends so I chose to talk them together for our first interview. Of the three, only Flix and Eugne had confessed to participating in the genocide; Paul maintained his innocence during the genocide and so had not confessed. At 52, Flix was the oldest of the three but seemed, by far, the most scared and nervous during the interview. With his slouched shoulders, bowed head, and eyes darting left to right to avoid eye contact, Flix gave the impression of a frightened boy steeling himself for a blow he knew would come at any moment. Flixs nervous demeanor contrasted sharply with Pauls garrulousness and Eugnes detached air. The group interview did not last long, mostly because Paul, with his outgoing personality, quickly took over the interview and answered all the questions himself, while Eugne and Flix seemed content to remain silent. What did come up in this brief interview, however, was insight into the meaning of friendship the three shared before the genocide. Q. Did you three know each other before the genocide? Since when? Yes, we knew each other because we had been neighbors for a long time. Q. You went to the same primary school? No. Q. Before the genocide, how much time did you spend together? We were neighbors, we would share everything. Despite some being older than others, we were together a long time until the period of the genocide and the war. Q. How did you become friends? We lived in the same secteur, almost the same neighborhood. We would help each other. All that. Its that no one wished anything bad to come to
This demographic profile also fits the Joiners from my northern research site in Ruhengeri. Moreover, these data are consistent with Strauss (2004) national random sample of confessed and sentenced killers.
17

- 11 -

anyone else. Q. Can you give me an example where you needed help and the others helped you? For example, helping the other with work in the field, cultivating together. Another example: Helping each other at the time of a wedding where there was someone who would be receiving a cow. Another example: Helping each other take the sick to the hospital [Groupe #1/1]. This description of what constituted the friendship between the three is consistent with others I queried on the topic, Joiners and non-Joiners alike. Friendship was facilitated by physical proximity which enabled frequent interaction. Friendships were constituted by shared activities and mutual gestures of help and support. Friendship did not lead the three to make the same choices during the genocide, however. This difference is not surprising given the trios radically different positions in the community when the genocide began. As conseiller, Paul had been in a position of authority and was even able to use his former position to help save Tutsi. His ability to help those in danger rested not only on his refusal to turn over the secteur stamp and registry to Jude, but also on his friendships and relationships to Tutsi of Ngali and surrounding secteurs. Q. How did the registry aid you in helping the Tutsi at that time so that they could go anywhere? To put it briefly, Jude demanded that I give him the secteur stamp from the very beginning and I refused. I gave papers to Tutsi that said that this Tutsi had lost his identity card and that he was going to obtain another one after the war. On this paper, I would write that the person was Hutu and with this paper, he could travel anywhere except that there were times when they would kill someone because of their physical characteristics. Q. How many people did you give this paper to? I dont remember very well. It was at least between 35-40 people. I would even give these papers to people from other secteurs like the people of secteur [names 4 secteurs] in the commune of Kigoma. Q. How did the people in Ngali find you? They would send their friends to my house to ask me to write a paper for a certain Tutsi but the others would come by during the night. Q. How did the people from the other secteurs know that you were giving this paper to Tutsi? I had friends in the other secteurs who were married to Tutsi who asked me for advice. And when I gave them a paper, they would go tell others what happened. [Paul, __] Pauls relations benefited the Tutsi of Ngali and even those of surrounding secteurs. One of the people who received such a paper was Pauls friend, Eugne. Eugne (Joiner, Ngali) Eugne was a young man of 25 when the genocide began in 1994. He had

- 12 -

married only two years before and had one child. Like most people in Ngali, he experienced no problems when the civil war started in Rwanda in 1990. The war was far away and life remained calm in Ngali. After the assassination of the president, however, people began to threaten and chase Eugne because he was among those who had to be killed. Eugnes prior relations with a group of close friends shaped his range of options in unexpected ways. First, Eugne was able to obtain the stamped document from Paul, which allowed him to move about freely. Rather than use this document to flee the area, however, Eugne used it to join the Interahamwe, in order, he said, to save himself. By joining the Interahamwe, Eugne was not placing himself in the hands of strangers, but re-joining a group of longtime friends. Q. Why did you decide to confess? It was because I saw the death of a lot of people. Q. You confessed to having done what? I was witness to everything I saw and by whom. Q. You yourself killed no one? Me, I didnt kill anyone. Q. Even though you were among a group of Interahamwe? Yes. Q. What were you doing in this group if you didnt participate in the killings? Me, I was an observer. Q. What was your role, your job, as observer? We were very numerous. It was the young men I had known for a long time. I was among the young people during the war and I saw everything that they did and I had no role. I was with them in order to save myself. Q. The other young people in your group were your close friends? Yes. Q. No one in your group wanted you to participate in their activities? Yes, there was one who gave me a person to kill but I refused to kill her [Eugne, #3/4]. The extent of Eugnes actual participation may not have been only observer. In his letter of confession to the prosecutor in Gitarama, Eugne details the killings of five specific victims (as is required for confessions) and specifies his own participation as follows: I was always with the group of Interahamwe who would go hunt Tutsi. I participated in the genocide because there were times when I would give Tutsi to the group of Interahamwe and at the roadblock. Whether his participation was direct or indirect, it is clear that what influenced Eugnes survival strategy was the history of relations he had with those in the killing group. Q. How many people did your group kill? More than 10 people? It was a lot. More than 10 people. It was my group that killed nearly everyone in our secteur.

- 13 -

Q. How did your group choose their victims? The group killed Tutsi. Q. Did you know all your groups victims? Yes, they were our neighbors. Because you could know all the people in your secteur. Q. And all the members of your group of killers were also from Ngali? Yes, they were from Ngali secteur. Q. Over how many days did your group kill? I was with the group for a month. They went to the Congo, and as for me, I stayed in Ngali. Q. When did your group form as a group? Before the war,18 we were a group and we continued this group even during the war. Q. A group of what? friends? killers? bandits? Before the war, we were friends. We were in business selling food and small livestock [Eugne, #3/4]. Eugnes responses to various questions highlight both the social proximity between killers and victims as well between killers in the same group. Because Eugne and his friends constituted a group before the genocide, they did not have to create a new group from scratch during the genocide; they merely had to maintain what had been in place before, even if this meant allowing a Tutsi into the Interahamwe. Undoubtedly, Eugnes friends knew he was Tutsi, but let him pass as a Hutu during the duration of the genocidal violence in Ngali. As with many Interahamwe, prior connections to Tutsi presented Joiners with stark moral dilemmas. Eugne had grown up with his mother and maternal grandparents because his father was often away in Bujumbura. While growing up, Eugnes maternal uncle was a central figure in his life, doing everything for the young boy including paying his school fees. During the genocide, Eugnes uncle went into hiding. Even after joining the Interahamwe, Eugne tried to help his uncle by bringing him food. Eugnes efforts were in vain as a group of killers from an adjoining secteur found the man and killed him. When I asked Eugne how he heard the news of his uncles death, he replied simply: I was living in the cellule M and bad news didnt hide itself for long. They told me that the people had just killed my uncle [Eugne, #2/4]. Despite his inability to save his uncle, Eugne was at least spared the ordeal of having to watch his own group kill the man. When I asked Eugne what he would have done had Jude or one of Judes lieutenants ordered him to kill his uncle himself, his answer was surprisingly frank. After hearing the question, Eugne paused for a moment, then said in an even, matter-of-fact tone: Its hard to say. Maybe I would have killed him, to save my life [Eugne, #2/4]. As Eugnes case clearly illustrates, the density of ties which linked people in multiple ways before the genocide also pulled people in multiple and often unexpected
The majority of my sample in both research communities referred to the genocide using the Kinyarwandan word intambara which means war despite there being two different ways in Kinyarwanda to refer to genocide.
18

- 14 -

directions during the genocide. The configuration of these ties is structured how people saw their options. Michel (Joiner, Ngali) Eugne was by no means the only Joiner trying to save Tutsi even while participating in the hunt and mass murder of other Tutsi. Michel was a Joiner who told us how he saved several Tutsi by hiding them in his house. Q. The last time you told us that you saved four people. Tell us how you decided to hide these four people. I saw that the killing was terrible and I saw especially that our neighbors were suffering, that I had to something good. I give you the example of the lady GN. Between her house and mine, there is a sister house. As we had shared everything before, I decided to save her. Q. Did GN ask you to save her? She did not ask me to save her. She only arrived at our house and we agreed to hide her and to share with her what we had at the time until the end of the war. Q. Who were you with at your house? My sister [], my little brothers, []. Except that [one of my brothers] had his own house and we hid them together. Q. Did you hesitate to hide this woman at first? We did not hesitate to hide her except that each time the killers came to search for people to kill. But we did everything possible to protect her [Michel, #__]. Michels rescue activities did not keep him from participating in the murder of another Tutsi, however. This victim was a stranger to Ngali and had stopped at Michels house to ask for directions to another neighbors house. Michel describes his actual involvement in this mans murder in ambiguous terms. Q. What happened next at your house? after this mans arrival? Once he got there, he asked me to show him the path that leads toward ENs house. I made him stay a while at my house [i.e., dally] and after a few minutes, a man named KK arrived at my house and asked him to follow him. This man followed KK to the road block where they were working. Q. Who was this man named KK? He was the leader of the road block. Q. What did he want with this man? He wanted to have him killed and he was killed at the road block. Q. Were you able to see the killing of the man from your house? No, because between the road block and my house, there is a forest. Q. How did you know this man was killed at the road block?

- 15 -

The killers passed by near my house saying they had killed him. Q. How did KK know that this man was at your house? I dont know. Q. Why did KK and the others want to kill the man who was at your house? When KK arrived at my house, he started to ask him lots of questions and maybe he noticed that this man was among those people who they were supposed to bring to the road block. Q. Did you know that KK was going to kill this man when he took him to the road block? Yes, I thought of his death because there were some others who were killed like that except that I had nothing to do for him. Q. Did you try to keep KK from taking this man from your house? Yes, I tried to prevent him but he didnt listen to me. Afterwards, I ran into EN [the woman whose house the victim had been looking for] who had me imprisoned and I told her that KK did not want to let this man go. Q. What did you say to KK in trying to keep him from taking this man to the road block? After talking to this man, KK asked me what this man was doing at my house. I told him that he came to ask me the path which leads to ENs family. He told me he was going to take him to the road block and I asked him to let him go to ENs house before taking him to the road block. KK refused and asked the man to follow him. Q. Was KK alone or was he with some others? Yes, he was alone. Q. Was he armed? I dont remember if he was armed [Michel, #__]. In his letter of confession, Michel paints a different picture. In this version of the story, Michel not only helped escort the man to the road block but also had a direct hand in killing him. Even in his interview version, however, Michels complicity is evident, particularly in the statement where he admits to delaying the man at his house. That KK shows up a few moments later would imply some kind of coordination. Also, Michels effort at preventing KK from taking the man to the road block was hardly a sincere effort at trying to save the stranger since all Michel suggested was taking the man first to the mans friends house and then to the road block, where the man would have been killed anyway. Whatever the extent of Michels actual involvement in the mans death, the difference between the man and the four Tutsi Michel was hiding was that Michel knew the latter but not the former. It was, after all, Michels own neighbors that he saw suffering when the genocide began. Michel had no such ties to the strange man, and it seems that the mans outsider statushis lack of any ties to the communitymade him vulnerable to targeting by KK and unworthy of protection by Michel. Stefan (Joiner, Kimanzi) While ties between Joiners could create new options for some, like Eugne, ties

- 16 -

could also simply lead other Joiners unwittingly into participating in the genocide, as was the case with a Joiner from Kimanzi. Stefan was a slight man when we met him the prison in Ruhengeri in 2004. His face was hollow, with deep-set eyes and sunken cheeks, making him look much older than his age. In 1991, the year when he took part in the mass killings of Tutsi in Ruhengeri, he would have been 27 years old, a married man with 3 children. He was a close neighbor of the victim, the sole Tutsi living in the area. As neighbors, the two had shared beers, conversation, and chores over the years. They were friends according to Stefan, in addition to being neighbors. Like Michel, Stefans describes his involvement in very ambiguous terms, wavering between having been mere witness to the murder of his neighbor to having been a direct participant in the mans murder. Q. Did you see the killings of your neighbors? I saw the death of GK and it was he whom we killed. Q. How was he killed? We were by the road, for example [names people in the group]. We were hearing that at [my neighbors] house, there were ibyitso of the Tutsi. We arranged the day to go attack his house. Me and three others, we went there as well. When we arrived there, we searched for the ibyitso but there werent any. It was [my neighbors] son, GK. [One in my group] said to GK: Its you Tutsi who are causing all these problems. And he killed him with a machete [Stefan, #1/5]. What emerges from Stefans accounts of his neighbors murder is the importance of Stefans prior interactions in a group of peers. Stefan and his group received and processed information as a group and decided as a group what to do about the rumors they were hearing about Stefans neighbor hiding ibyitso. These determinations overrode Stefans doubts that his neighbor, his erstwhile friend, was hiding ibyitso. Q. At that time, how did the people distinguish between the Tutsi who were true ibyitso and those Tutsi who were not? The community made no distinction between one kind of Tutsi and the other. At that time, all Tutsi were ibyitso. Q. And thats what you believed at that time? People differ from one another. For me, like my neighbor, I didnt think that he was an icyitso but the others had just come to tell me that he was an icyitso whatever that was [Stefan, #2/5]. Stefans last remark is quite telling. Despite his own (individual) doubts that his neighbor and friend was really an icyitso, he nonetheless acceded to the groups determination that the man must be icyitso because of the rumors and edicts passed down from local authorities. Like Eugnes group, Stefans was a group of neighbors, not strangers. Like most residents, these neighbors had known each other for a long time. Many of them were also neighbors of the victim. The confluence of group interaction and authorities calls to root out the ibyitso hiding in their midst was powerful enough to override individual hesitations or doubts. Prior ties between Joiners helped bring these men together initially. By joining together as a group in pursuit of Tutsi and ibyitso, Joiners strengthened these prior ties.

- 17 -

Conclusion
There is very little evidence that participation emanated from either ethnic hatreds or ethnic fears. Rather, the evidence points to hatreds at the individual level that is, hatreds that were personal in nature, not ethnic. The evidence also points to fears of those in power, not of those in a particular ethnic group (i.e., Tutsi). There was clear reason to fear those in power for they did not hesitate to use force and threats of force to conscript participants and to dampen resistance. Thus, it was leaders and collaborators who presented the most immediate threat to Hutu and Tutsi alike. Following Granovetter, I argue instead that participation in the genocide was socially embedded in a set of densely-knit, overlapping ties. These relations were multiplex in that people were linked together across different roles (Portes 1998, 16, fn 6, quoting Boissevain 1974, 31-33). The same person might be ones neighbor, relative, and collaborator. The multiplex nature of these ties meant that people were often linked to each other in numerous ways. Before the genocide, these multiple linkages could be a help; during the genocide, they were often a liability. In each community, local leaders derived their power and authority from connections they had to those higher up in the government or political party chain. (To be sure, these two networks often converged.) Immediate patrons were generally the bourgmestre of the commune, but could include military soldiers or other agents of the state. Such connections enabled leaders and collaborators to wield absolute power in their immediate communities. Because power was centered on one undisputed leader or leaders in each community, proximity to a center delimited a persons options, by making some options more readily available than others. One avenue that close proximity cut off was the option of retreat into silence or ignorance. This was the option that many people took. One of the most common statements from people (even Joiners) was I stayed at home or I stayed in the house when describing what they did during the period of the war and genocide. For those situated far from the center of power, staying at home was not only an available option, it was perhaps the option that made the most sense given the level of insecurity and uncertainty that pervaded each community, particularly in the aftermath of an attack by the RPF or the assassination of the president. When those situated closest to leaders tried to opt out, they were accused of being ibyitso, and, as a result, became targets themselves. Ties based on family could thus be the most pernicious in terms of constraining the range of options for those otherwise disinclined to join the violence. Family ties could, at the same time, be advantageous to members who sought to profit from the genocide (i.e., denouncers). Vertical ties, however, were not the only important set of relations that mattered in terms of structuring Joiners participation. Horizontal ties were also important. First, they facilitated Joiners coming together as a group. Prior ties between Joiners meant that Joiners already had a history of collaborating and working together in groups and as groups. When crisis moments came, prior ties made it a natural step for Joiners to come together as a group once again. Second, having facilitated Joiners coming together, horizontal ties then became sites for Joiners to make sense of their immediate situation and to decide what to do about it as a group. This process of making sense did not necessarily render previous ties to Tutsi friends and neighbors null and void, but the fact that Joiners tended to help only those with whom they shared previous ties19 underscores the power of Joiners ties to one another to override Joiners individual
Joiners who helped Tutsi, for example, tended to help Tutsi they knew and liked. Rescuers, by contrast, (i.e., those who helped to save people without ever participating in the genocide), saved strangers and friends alike.
19

- 18 -

hesitations to hurt Tutsi friends and neighbors. In other words, it is precisely because Joiners retained their sense of ties and loyalty to their Tutsi neighbors and friends that some other mechanism was needed to override individual Joiners hesitations at targeting these same neighbors and friends. Finally, horizontal ties were important because they created pathways to participation that would not have otherwise existed. During a genocide of Tutsi, Tutsi are not supposed to be able to opt out of being victims to become killers, yet that is exactly what Eugne and the other Tutsi he mentioned did. Money was a factor, to be sure, (the payments made to Jude), but so too was Eugnes previous relationships to the other members in the Interahamwe. Absent these previous ties, Eugne would have had to pursue a different strategy to save himself. So was the case for other Joiners. For absent the ties between them, other Joiners, too, may have been able to follow a different path. They may well have stayed at home like so many others did, leaving the work of genocide to others.

References
Betrand, Jordane. 2000. Rwanda: Le pige de l'histoire. Paris: Karthala. Coleman, James S. 1988. Social capital in the creation of human capital. American Journal of Sociology 94 (Supplement):S95-S120. de Figueiredo, Jr., Rui J.P., and Barry R. Weingast. 1999. The rationality of fear: Political opportunism and ethnic conflict. In Civil wars, insecurity, and intervention, edited by Barbara F. Walter and Jack Snyder. New York: Columbia University Press. Des Forges, Alison. 1999. Leave none to tell the story. New York: Human Rights Watch & Fdration internationale des ligues des droits de l'homme. Gagnon, Jr., V.P. 2004. The myth of ethnic war: Serbia and Croatia in the 1990s. Ithaca: Cornell University Press. Granovetter, Mark. 1985. Economic action and social structure: The problem of embeddedness. American Journal of Sociology 91 (3):481-510. Granovetter, Mark S. 1973. The strength of weak ties. American Journal of Sociology 78 (6):1360-80. Hardin, Russell. 1995. One for all. Princeton: Princeton University Press. Hatzfeld, Jean. 2003. Une saison des machettes. Paris: Seuil. ________. 2000. Dans le nu de la vie: Rcits des marais rwandais. Paris: ditions du Seuil. Jefremovas, Villia. 2002. Brickyards to graveyards: From production to genocide in Rwanda. Albany: State University of New York Press. Kaplan, Robert D. 1994. The coming anarchy. Atlantic Monthly, February, 44-73. Kaufman, Stuart J. 2001. Modern hatreds: The symbolic politics of ethnic war. Ithaca: Cornell University Press.

- 19 -

Lemarchand, Ren. 1970. Rwanda and Burundi. New York: Praeger Publishers. Mamdani, Mahmood. 2001. When victims turn killers. Princeton: Princeton University Press. Mueller, John. 2000. The banality of "ethnic war". International Security 25 (1):42-70. Newbury, Catharine. 1988. The cohesion of oppression: Clientship and ethnicity in Rwanda, 1860-1960. New York: Columbia University Press. Portes, Alejandro. 1998. Social capital: Its origins and applications in modern sociology. Annual Review of Sociology 24:1-24. Posen, Barry R. 1993. The security dilemma and ethnic conflict. Survival 35 (1):27-47. Prunier, Grard. 1995. The Rwanda crisis: History of a genocide. New York: Columbia University Press. Putnam, Robert D. 1993. Making democracy work. Princeton: Princeton University Press. Straus, Scott. 2000. How many perpetrators were there in the Rwandan genocide? An estimate. Journal of Genocide Research 6 (1):85-98. Straus, Scott Alexander. 2004. The order of genocide: Race, power, and war in Rwanda, Political Science, University of California, Berkeley. Varshney, Ashutosh. 1997. Postmodernism, civic engagement, and ethnic conflict: A passage to India. Comparative Politics 30 (1):1-20. Varshney, Ashutosh. 2002. Ethnic conflict and civic life: Hindus and Muslims in India. WNew Haven: Yale University Press. Wagner, Michelle D. 1998. All the bourgemestre's men. Africa Today 45:25-36.

- 20 -