Rwandans take history seriously. Hutu who killed Tutsi did so for many reasons, but beneath the individual motivations lay a common fear rooted in firmly held but mistaken ideas of the Rwandan past. Organizers of the genocide, who had themselves grown up with these distortions of history, skillfully exploited misconceptions about who the Tutsi were, where they had come from, and what they had done in the past. From these elements, they fueled the fear and hatred that made genocide imaginable. Abroad, the policy-makers who decided what to do—or not do—about the genocide and the journalists who reported on it often worked from ideas that were wrong and out-dated. To understand how some Rwandans could carry out a genocide and how the rest of the world could turn away from it, we must begin with history.

Forerunners of the people who are now known as Hutu and Tutsi settled the region over a period of two thousand years. Originally organized in small groups based on lineage or on loyalty to an outstanding leader, they joined in building the complex state of Rwanda. They developed a single and highly sophisticated language, Kinyarwanda, crafted a common set of religious and philosophical beliefs, and created a culture which valued song, dance, poetry, and rhetoric. They celebrated the same heroes: even during the genocide, the killers and their intended victims sang of some of the same leaders from the Rwandan past.1

In early times, as now, most people in the region were cultivators who also raised small stock and occasionally a few cattle. A far smaller number of people scorned cultivation and depended on large herds of cattle for their livelihood. Cultivators and pastoralists lived interspersed in most areas, although the cool, wet highlands of the north had few pastoralists and the drier, hotter east had more. With fertile soil and regular rainfall, the region was productive and population grew to a point where Rwanda was in 1994 the most densely populated nation on the African continent.

When Rwanda emerged as a major state in the eighteenth century, its rulers measured their power in the number of their subjects and counted their wealth in the number of their cattle. The two were usually related. Giving or temporarily granting cattle was a way of winning supporters; a large number of supporters helped to win cattle, both in conflicts with other members of the elite and inadventures abroad. But not all cattle-owners held state positions. The pastoralists known as Bagogwe, clustered in the northwest, and those called Bahima, located in the northeast, sought to avoid state power rather than to share in it. Conversely, not all members of the elite were born rich in cattle, although those lacking such wealth ordinarily acquired it along with power. Cultivators skilled in making war and able to mobilize large groups of followers rose to importance through the military system, particularly under the late nineteenth century ruler Rwabugiri, who brought Rwanda to the height of its power. In its drive to expand, Rwanda attacked neighboring peoples regardless of whether they were pastoralists or cultivators and regardless of whether they were organized in lineages or in states.2

Rwandan institutions were shaped by both pastoralists and cultivators. Although the power of the ruler derived from control over the military and over cattle, his authority was buttressed also by rituals firmly rooted in agricultural practices.3 By the end of the nineteenth century, the ruler governed the central regions closely through multiple hierarchies of competing officials who administered men, cattle, pasturage, and agricultural land. He exercised a looser kind of suzerainty over other areas, particularly on the periphery, which were dominated by powerful lineage groups, some of them pastoralists, some cultivators. In addition, he tolerated the existence of several small states within the boundaries of Rwanda, usually because their rulers were thought to control rainfall, crop pests, or some other aspect of agricultural productivity important for Rwanda as a whole. The late President Habyarimana and his circle counted themselves as the proud contemporary representatives of Bushiru, the largest such state within Rwanda at the beginning of the colonial era.

As the Rwandan state grew in strength and sophistication, the governing elite became more clearly defined and its members, like powerful people in most societies, began to think of themselves as superior to ordinary people. The word “Tutsi,” which apparently first described the status of an individual—a person rich in cattle—became the term that referred to the elite group as a whole and the word “Hutu”—meaning originally a subordinate or follower of a more powerful person—came to refer to the mass of the ordinary people. The identification of Tutsi pastoralists as power-holders and of Hutu cultivators as subjects wasbecoming general when Europeans first arrived in Rwanda at the turn of the century, but it was not yet completely fixed throughout the country. Rulers of small states embedded in the larger nation, important lineage heads and some power-holders within the central state hierarchy exercised authority even though they were people who would today be called “Hutu.”

Most people married within the occupational group in which they had been raised. This practice created a shared gene pool within each group, which meant that over generations pastoralists came to look more like other pastoralists—tall, thin and narrow-featured—and cultivators like other cultivators—shorter, stronger, and with broader features. Within each group there were also sub-groups, the result of some distant common ancestry or of more recent patterns of marriage. Thus among pastoralists, some whose ancestors had arrived centuries ago were distinctly shorter, plumper, and redder-skinned than the taller and blacker-skinned descendants of nineteenth-century immigrants. Cultivators, who were relatively sedentary and chose mates from areas close to home, often exhibited traits characteristic of their places of origin: those from the south, for example, were generally shorter and slighter than those from the north central region.

Although it was not usual, Hutu and Tutsi sometimes intermarried. The practice declined in the late nineteenth and early twentieth centuries as the gap widened between Tutsi elite and Hutu commoners, but rose again after Tutsi lost power in the 1959 revolution. With the increase in mixed marriages in recent decades, it has become more difficult to know a person’s group affiliation simply by looking at him or her. Some people look both “Hutu” and “Tutsi” at the same time. In addition, some people who exhibit the traits characteristic of one group might in fact belong to the other because children of mixed marriages took the category of their fathers, but might actually look like their mothers.4 During the genocide some persons who were legally Hutu were killed as Tutsi because they looked Tutsi. According to one witness, Hutu relatives of Col. Tharcisse Renzaho, the prefect of the city of Kigali, were killed at a barrier after having been mistaken for Tutsi.5

The Twa, a people clearly differentiated from Hutu and Tutsi, formed the smallest component of the Rwandan population, approximately 1 percent of the total before the genocide. Originally forest dwellers who lived by hunting and gathering, Twa had in recent decades moved closer to Hutu and Tutsi, working aspotters, laborers, or servants. Physically distinguishable by such features as their smaller size, Twa also used to speak a distinctive form of Kinyarwanda. While the boundary between Hutu and Tutsi was flexible and permeable before the colonial era, that separating the Twa from both groups was far more rigid. Hutu and Tutsi shunned marriage with Twa and used to refuse even to share food or drink with them. During the genocide, some Twa were killed and others became killers. Because Twa are so few in number and because data concerning them are so limited, this study does not examine their role.

The Germans, who established a colonial administration at the turn of the century, and the Belgians who replaced them after the First World War, ended the occasional open warfare that had taken place within Rwanda and between Rwanda and its neighbors. Both Germans and Belgians sought to rule Rwanda with the least cost and the most profit. Making use of the impressive indigenous state was the obvious way to do so, but the colonialists found its complexities troublesome. The multiple hierarchies which had allowed the ruler to maximize his control by playing off rival officials now permitted both ruler and his subordinates to evade control by the colonialists. The dense administration within central Rwanda—with the least important representatives of the ruler sometimes governing only a few hundred people—required a relatively high proportion of local goods and labor for its support. The colonialists preferred to have these resources at their own disposal, to cover their expenses and to pay the costs of building an infrastructure to link Rwanda to the world economy. At the same time, the Belgians saw the autonomous enclaves, where central control was light, as anomalies potentially disruptive of good order.

In the 1920s, the Belgians began to alter the Rwandan state in the name of administative efficiency. Always professing an intention to keep the essential elements of the system intact, they eliminated the competing hierarchies and regrouped the units of administration into “chiefdoms” and “sub-chiefdoms” of uniform size.They used force to install state officials in the autonomous enclaves, destroying the power of the heads of lineages and of local small states. They fixed and made uniform the goods and services that local officials could demand, thus—they thought—reducing the burdens on the population.

Rwandan officials were not helpless pawns but rather real players in the game of administrative reform. Politically astute, they understood how to evade the intent of European orders even while apparently conforming to them. Chiefs and sub-chiefs seemed to accept the reduction in numbers of officials, but in fact kept on using unofficial representatives out on the hills who continued living off the localpeople. As a result, the density of administration and consequent customary burdens on the people diminished little, if at all, in the central part of the country, while in the north and southwest, they actually increased because of the installation of resident officials. At the same time, the chiefs and sub-chiefs—and later other administrative agents—enforced a series of wholly new demands imposed by the colonialists as part of their effort to integrate Rwanda into the world economy. They often found ways to turn these new requirements, such as building roads or planting cash crops, to their personal profit.

The elite profited not just from direct European backing but also from the indirect and unintended consequences of the administrative changes. Under the old system of multiple officials, power-holders ordinarily limited demands on subordinates, knowing that those who felt unreasonably exploited could seek protection from rivals or could move elsewhere, even clearing new land in the forest, if need be, to escape exactions. In the 1920s and 1930s, the Belgians made it far harder for the weak to escape repressive officials; not only did they eliminate the multiple hierarchies but they also restricted changes in residence from one region to another and they prohibited new settlement in the forests. The one avenue of escape still possible was migration abroad and thousands took that route beginning in the 1920s. But those who preferred not to leave Rwanda had little choice but to submit to increased exploitation of officials now freed from the constraints that once limited their demands.

European administrators generally overlooked the abuses of those officials who got the taxes collected, the roads built, and the coffee planted. They established European-style courts which they expected would protect the ordinary people, but they usually did not. The judges saw themselves as defenders of the elite, not the masses.

At the same time that the Belgians enabled the officials to demand more from the people, they decreed that Tutsi alone should be officials. They systematically removed Hutu6 from positions of power and they excluded them from higher education, which was meant mostly as preparation for careers in the administration. Thus they imposed a Tutsi monopoly of public life not just for the 1920s and 1930s, but for the next generation as well. The only Hutu to escape relegation to the laboring masses were those few permitted to study in religious seminaries.

By assuring a Tutsi monopoly of power, the Belgians set the stage for future conflict in Rwanda. Such was not their intent. They were not implementing a“divide and rule” strategy so much as they were just putting into effect the racist convictions common to most early twentieth century Europeans. They believed Tutsi, Hutu, and Twa were three distinct, long-existent and internally coherent blocks of people, the local representatives of three major population groups, the Ethiopid, Bantu and Pygmoid. Unclear whether these were races, tribes, or language groups, the Europeans were nonetheless certain that the Tutsi were superior to the Hutu and the Hutu superior to the Twa—just as they knew themselves to be superior to all three. Because Europeans thought that the Tutsi looked more like themselves than did other Rwandans, they found it reasonable to suppose them closer to Europeans in the evolutionary hierarchy and hence closer to them in ability. Believing the Tutsi to be more capable, they found it logical for the Tutsi to rule Hutu and Twa just as it was reasonable for Europeans to rule Africans. Unaware of the “Hutu” contribution to building Rwanda, the Europeans saw only that the ruler of this impressive state and many of his immediate entourage were Tutsi, which led them to assume that the complex institutions had been created exclusively by Tutsi.

Not surprisingly, Tutsi welcomed these ideas about their superiority, which coincided with their own beliefs. In the early years of colonial rule, Rwandan poets and historians, particularly those from the milieu of the court, resisted providing Europeans with information about the Rwandan past. But as they became aware of European favoritism for the Tutsi in the late 1920s and early 1930s, they saw the advantage in providing information that would reinforce this predisposition. They supplied data to the European clergy and academics who produced the first written histories of Rwanda. The collaboration resulted in a sophisticated and convincing but inaccurate history that simultaneously served Tutsi interests and validated European assumptions. According to these accounts, the Twa hunters and gatherers were the first and indigenous residents of the area. The somewhat more advanced Hutu cultivators then arrived to clear the forest and displace the Twa. Next, the capable, if ruthless, Tutsi descended from the north and used their superior political and military abilities to conquer the far more numerous but less intelligent Hutu. This mythical history drew on and made concrete the “Hamitic hypothesis,” the then-fashionable theory that a superior, “Caucasoid” race from northeastern Africa was responsible for all signs of true civilization in “Black” Africa. This distorted version of the past told more about the intellectual atmosphere of Europe in the 1920s than about the early history of Rwanda. Packaged in Europe, it was returned to Rwanda where it was disseminated through the schools and seminaries. So great was Rwandan respect for European education that this faulty history was accepted by the Hutu, who stood to suffer from it, as well as by the Tutsi who helped to create it and were bound to profit from it. People of both groups learned to thinkof the Tutsi as the winners and the Hutu as the losers in every great contest in Rwandan history.

The polished product of early Rwando-European collaboration stood unchallenged until the 1960s when a new generation of scholars, foreign and Rwandan, began questioning some of its basic assumptions.7 They persuaded other scholars to accept a new version of Rwandan history that demonstrated a more balanced participation of Hutu and Tutsi in creating the state, but they had less success in disseminating their ideas outside university circles. Even in the 1990s, many Rwandans and foreigners continued to accept the erroneous history formulated in the 1920s and 1930s.

Once the Belgians had decided to limit administrative posts and higher education to the Tutsi, they were faced with the challenge of deciding exactly who was Tutsi. Physical characteristics identified some, but not for all. Because group affiliation was supposedly inherited, genealogy provided the best guide to a person’s status, but tracing genealogies was time-consuming and could also be inaccurate, given that individuals could change category as their fortunes rose or fell. The Belgians decided that the most efficient procedure was simply to register everyone, noting their group affiliation in writing, once and for all. All Rwandans born subsequently would also be registered as Tutsi, Hutu, or Twa at the time of their birth. The system was put into effect in the 1930s, with each Rwandan asked to declare his group identity.8 Some 15 percent of the population declared themselves Tutsi, approximately 84 percent said they were Hutu, and the remaining 1 percent said they were Twa. This information was entered into records at the local government office and indicated on identity cards which adult Rwandans were then obliged to carry. The establishment of written registration did not completely end changes in group affiliation. In this early period Hutu who discovered the advantages of being Tutsi sometimes managed to become Tutsi even after the records had been established, just as others more recently have found waysto erase their Tutsi origins. But with official population registration, changing groups became more difficult.

The very recording of the ethnic groups in written form enhanced their importance and changed their character. No longer flexible and amorphous, the categories became so rigid and permanent that some contemporary Europeans began referring to them as “castes.” The ruling elite, most influenced by European ideas and the immediate beneficiaries of sharper demarcation from other Rwandans, increasingly stressed their separateness and their presumed superiority. Meanwhile Hutu, officially excluded from power, began to experience the solidarity of the oppressed.

Belgium continued its support for the Tutsi until the 1950s. Then, faced with the end of colonial rule and with pressure from the United Nations, which supervised the administration of Rwanda under the trusteeship system, the colonial administrators began to increase possibilities for Hutu to participate in public life. They named several Hutu to responsible positions in the administration, they began to admit more Hutu into secondary schools, and they conducted limited elections for advisory government councils. Hardly revolutionary, the changes were enough to frighten the Tutsi, yet not enough to satisfy the Hutu. With independence approaching, conservative Tutsi hoped to oust the Belgians before majority rule was installed. Radical Hutu, on the contrary, hoped to gain control of the political system before the colonialists withdrew.

The ruler who had been in power since 1931, Mutara Rudahigwa, had served to reassure all parties and to keep the situation calm. But he died unexpectedly in 19599 and was succeeded by a young half-brother, Kigeri Ndahindurwa, who appeared to be heavily influenced by the most conservative Tutsi group. Moderate parties that sought to organize across the Hutu-Tutsi divide lost ground as the Parmehutu (Parti du mouvement de l’émancipation des Bahutu), identified exclusively with Hutu, and the Union Nationale Rwandaise (UNAR), a royalist Tutsi party, gained in strength. In November 1959, several Tutsi assaulted a Hutu sub-chief. As the news of the incident spread, Hutu groups attacked Tutsi officials and the Tutsi responded with more violence. Several hundred people were killed before the Belgian administration restored order. The Belgians then replaced abouthalf the Tutsi local authorities by Hutu. With the help of many of these local administrators, the Parmehutu easily won the first elections in 1960 and 1961. In September 196l, some 80 percent of Rwandans voted to end the monarchy, thus confirming the proclamation of a republic the previous January 1961 by the Parmehutu-led government. These events became known as the “Hutu Revolution.”

In later years, and particularly during the genocide, Hutu politicians waved the flag of the revolution, knowing they would get an overwhelming response from their audiences. In fact the revolution was neither so heroic nor so dramatic as it was later presented. In their struggle for power, the Hutu were “helped” considerably by the Belgians, both politically and militarily. At the start, Hutu attacked power-holders and those related to them, leaving their ordinary Tutsi neighbors in peace. They usually sought to drive Tutsi away rather than to destroy them. The assailants cleared the north most completely, the area where Tutsi officials had been installed three decades before by the colonial administration. Many displaced Tutsi resettled elsewhere in Rwanda, particularly in the sparsely populated region known as Bugesera, but another 10,000 took the road to exile.

In 1961 some of these refugees began to attack Rwanda, an effort they would repeat ten times over the next six years. After these incursions, Hutu officials led reprisal attacks on Tutsi still within the country, accusing them of having aided the invaders—the same kind of charges often repeated at the time of the genocide.10 Only one of these attacks, that of late December 1963, posed a real threat to the new republic. But Hutu leaders used them all to bolster the sense of Hutu solidarity, to solidify their own control and to eradicate the last vestiges of respect for Tutsi authority. From these attacks they crafted the myth of the Hutu revolution as a long and courageous struggle against ruthless forces of repression. For them, the battle had been legitimate as well as brave: the Hutu, as the “great majority,” the “rubanda nyamwinshi,” had the right to rule over the minority. In their eyes, the ethnic majority was necessarily the same as the democratic majority.

At this time, Hutu politicians also established the link between “patriotism” and profit. In attacking the supposed enemies of the nation and the revolution, the Hutu stood to gain, both in the short term from goods pillaged and in the long term from lands appropriated from Tutsi who were driven away. Given the political and material gains from anti-Tutsi violence, officials and others had strong incentives to widen the circle of people targeted from the narrow group of former power-holders to all Tutsi. By 1967 when both the incursions and the attacks on Tutsi within Rwanda ended, Tutsi were at risk of attack for the simple fact of being Tutsi.During these years, some 20,000 Tutsi were killed and more than 300,000 were forced to flee abroad.11

The new republican government continued labeling all Rwandans as Hutu, Tutsi, or Twa, but the identity cards which had once served to guarantee privilege to Tutsi now served as a means to discriminate against them, both in employment and in education. Just as the new leaders maintained population registration, so they perpetuated the distorted concepts that had underlain the practice. Hutu used the ideas once prized by the Tutsi—ideas about Tutsi distinctiveness, foreign origins, and complete control over the Hutu—to justify the violence of the revolution and the discriminatory measures of the years after.

Following the revolution, the percentage of Tutsi in the Rwandan population declined sharply, partly because many had been massacred or fled, partly because some found ways to redefine themselves as Hutu. Said to represent 17.5 percent of the population in 1952, Tutsi were counted as only 8.4 percent of the total in 1991.12

Over a period of several years, the Parmehutu leaders, who were based in the south, eliminated Hutu rivals as well as the once powerful Tutsi and created what was in effect a single party state. By the end of the first decade of the republic, however, they were increasingly challenged by Hutu from the north who saw that all rhetoric about Hutu solidarity notwithstanding, the southerners were monopolizing the benefits of power. In the face of this growing split between Hutu of the north and Hutu of the south, “Public Safety Committees” and other groups began a campaign of intimidation and assaults on Tutsi in early 1973. Some attributed the attacks to southerners who hoped to minimize differences with northerners by reminding them of the common enemy; others laid them to northerners who hoped to create sufficient disorder to legitimate a coup d’état by the army, an institution dominated by northerners. Regardless of which group had initiated the campaign, the tactic was clear: seek to resolve differences among Hutu at the expense of the Tutsi.

In July 1973, General Juvénal Habyarimana, the most senior officer in the army, took power, promising to restore order and national unity. He established the second republic in what was at the time a bloodless coup, although some fifty ofthe most prominent leaders of the first republic subsequently were executed or died in prison.

Two years after the coup, in 1975, Habyarimana made Rwanda officially a single-party state under the National Revolutionary Movement for Development (Mouvement Révolutionnaire National pour le Développement, MRND).13 All Rwandans of whatever age were automatically members of the party. Over the years, Habyarimana constructed a cohesive monolith, with himself as president of the republic and president of the party and, at each level below him, the relevant government official simultaneously heading the corresponding level of the party.

At this time, Rwanda was divided into ten prefectures,14 each of which included sub-prefectures, administrative units without much political importance. Below them were the communes, the essential building blocks of the administration. Numbering 145 in 1991, the communes ranged in population from less than 30,000 for the smallest to over 100,000 for the largest, with most counting between 40,000 and 50,000 residents. The head of the commune, the burgomaster, of course ranked below the prefect or sub-prefect, but he exercised more immediate and pervasive power over the ordinary people than did his superiors. In a style that harked back to the pre-colonial and the colonial era, the burgomaster held court one or more times a week, receiving the ordinary people who brought him their grievances or who came to give thanks for help received. He determined the use of land that belonged to the commune or was temporarily under its control. He mediated conflicts over property, settled family disputes, found places in secondary school, dispensed political advice, and even judged a substantial number of cases that in principle should have been taken to court. In accord with the communal council, he hired and fired the employees of the commune, including the communal policemen who were at his command, and he also intervened in personnel decisions of local schools, health centers, and development projects, although sometimes the presence of expatriates on project staffs limited his influence in this domain. The ultimate authority at the local level, he was clearly and directly the president’s manout on the hills. Although nominally responsible to the minister of the interior, the burgomasters were named by Habyarimana and removed by him. All were known to him and some were very close to him personally.

The communes were divided into sectors, each of which had a population of some 5,000 people. The sectors were represented by elected councilors who together formed the communal council that supposedly advised the burgomaster, but more often simply implemented his decisions. The sector was in turn composed of cells, each of which grouped together approximately 1,000 people. The cell had an elected committee of five persons, headed by a responsable (cell head), who were charged more with executing orders from above than with representing the views from below. That small part of the population employed in urban salaried jobs participated in the party at their place of work, where the work unit was also a party cell.

This intensive administration had two objectives: control and mobilization. The control was implemented not just by the high ratio of officials to ordinary people but also by regulations governing population registration and movement. The Habyarimana government continued the use of identity cards and also required people moving from one location to another to register with the local authorities. Each commune submitted monthly, quarterly, and yearly reports of births, deaths, and movement into and out of the commune. The burgomaster kept agents of the secret service informed of any suspicious persons seen in his district. In his first months in office, Habyarimana ordered important government employees with master’s degrees or higher to take military training, apparently with the intention of providing one more channel for instilling habits of obedience to orders.

The mobilization of the population aimed at first towards building the economic infrastructure and improving conditions for agriculture. Exploiting the practice of unpaid, communal labor imposed by the colonial administration, the MRND required the population to do umuganda, work for the public good, such as repairing roads, digging anti-erosion ditches, or clearing the brush. Umuganda was supervised by the nyumbakumi, a neighborhood leader in charge of a group of ten households, who had the power to fine those who failed to appear for the communal work sessions.

Once the MRND was firmly established, mobilization took on an added aspect: glorifying the party and its head. In addition to the work days, people were obliged to participate in weekly sessions of animation, propaganda meetings leavened with poetry, music, and dance created to honor Habyarimana and the MRND. Propaganda teams of singers and dancers vied for honors in regular competitions, often dressed in fine costumes bought by contributions from the party faithful.Rwandans often proclaimed their loyalty to Habyarimana, wore his image on portrait pins, and posted his picture in their houses or places of business.

The Army, the Church and the Akazu

As head of the army, Habyarimana had the allegiance of some 7,000 troops of the Rwandan Armed Forces (Forces Armées Rwandaises, FAR), about 1,200 of whom were part of the National Police (Gendarmerie). He was loyally supported especially by the elite units, made up largely of men from his home region: the Presidential Guard, estimated at between 1,000 and 1,300 troops, the paracommandos and the reconnaissance troops. He occasionally had to counter plots by other officers, however, including that attributed to Col. Alexis Kanyarengwe in 1980. Kanyarengwe, who had served as minister of interior, was forced to flee the country.

Habyarimana also enjoyed active support from the heads of the parastatal corporations that controlled public services like gas, water and electricity, or bus transport, and those that oversaw the production and marketing of cash crops. He knew he could count on the intellectual elite, including professors at the national university and heads of hospitals. To keep their posts, they would avoid criticizing him even if some declined to join in glorifying him. He could call on the heads of private enterprises to contribute materially and politically to his cause, knowing they needed his approval for the state concessions that made their businesses profitable.

He benefited enormously from the support of the hierarchy of the Catholic Church, which counted 62 percent of Rwandans among its adherents. The church, initially a pillar of support for the Tutsi elite, switched sides even before the colonial administration did and helped make the Hutu revolution. Although the majority of clergy, religious brothers, and sisters were Tutsi—some 70 percent according to one knowledgeable estimate—seven of the nine bishops in place at the start of the genocide were Hutu.15 The archbishop of Kigali, Mgr. Vincent Nsengiyumva, was an ardent supporter of the president, known for wearing Habyarimana’s portrait pin on his cassock while saying mass. He served as a member of the central committee of the MRND for many years and resigned only when church authorities insisted that he end his openly political role in 1985.

The various Protestant churches, representing 18 percent of the population, had no unified position towards Habyarimana, but the Anglican hierarchy and theBaptist church generally supported him. The president of the Presbyterian Church was a member of the prefectural committee of the MRND in Kibuye.

Both Catholic and Protestant clergy cooperated with officials by passing on state announcements from the pulpit and by serving on councils, particularly those that reviewed development projects at the prefectural or communal level.

One more link strengthened the connections from top to bottom of this highly structured system: the network of personal relations. Members of the elite who left home for positions in the capital or at the university maintained close ties to their communes of origin, where they had parents or other relatives. They visited home often and were the messengers of choice if some special order needed to be transmitted from the top to local officials. This practice existed long before Habyarimana took power—in December 1963, for example, ministers had gone home to organize the killings of Tutsi out on the hills, but he exploited it to maximum advantage, as did those who took over from him during the genocide.

The akazu, or “little house,” was a special circle within the larger network of personal connections that worked to support Habyarimana. It was composed mostly of the people of Habyarimana’s home region, with Madame Habyarimana and her relatives playing a major role. Some exercised authority openly, such as Protais Zigiranyirazo, who was once prefect of Ruhengeri, or Seraphin Rwabukumba, who headed a powerful enterprise, La Centrale, while others operated behind the scenes, such as Colonel Elie Sagatwa, who was Habyarimana’s private secretary. When necessary, this group drew on military officers, like Col. Théoneste Bagosora, Major Leonard Nkundiye, and Captain Pascal Simbikangwa, to ensure their continued hold on power.16 Christophe Mfizi, once close to Habyarimana and head of the national information service, denounced the activities of this group, which he called the “Zero Network.” In an August 15, 1992 public letter to the president resigning his membership in the MRND, he declared that the intimates surrounding Habyarimana had taken control of the state and were milking it for private benefit.17

Prosperity, Short-Lived and Superficial

At the head of what was taken to be an honest and energetic administration, Habyarimana attracted substantial foreign assistance in the 1970s and 1980s. Withsuch help, the government constructed an impressive infrastructure, particularly of roads and telephone and electric service. For the first decade, the economy did better than others in the same region, with a net increase in gross national product in relation to population, an achievement all the more remarkable given that Rwanda also had one of the highest rates of population growth on the continent.18 Donor nations applauded these accomplishments, regarding Rwanda as one of the few promising “models” in Africa. The expatriate experts who implemented the assistance projects in the country took great satisfaction not just in the results obtained but also in the personal ties that they developed with Rwandan counterparts.19

Some Rwandans were indeed getting rich: those who worked for the state directly, those employed by its offshoots, parastatal enterprises, and those who ran economic development projects controlled by state officials. State employees and the military also used access to preferential treatment to build profitable private businesses. But the prosperity was both fragile and superficial. The mass of the people stayed poor and faced the prospect of getting only poorer. More than 90 percent lived from cultivation and while the population grew, the amount of land did not. The land available to ordinary cultivators actually diminished in some regions as local officials appropriated fields for development projects and as members of the urban elite bought out the poor, establishing themselves as absentee landlords. According to a government study done in 1991, the richest 16 percent of landowners held 43 percent of the land, while the poorest households tried to eke out a living on holdings that ranged from one quarter to three-quarters of a hectare, or less than an acre of land.20 In the most densely populated regions, some young people could not marry because they could not find land and, according to custom, a man without land could not take a wife. This situation was so critical in Ngoma commune, Butare prefecture, that large numbers of young people were cohabiting and having children without marrying, a practice that broke dramatically with paststandards of behavior. Of the births registered in Ngoma, Butare prefecture, in January 1994, nearly 50 percent of the children had been born out of wedlock.21

At the end of the 1980s, coffee, which accounted for 75 percent of Rwanda’s foreign exchange, dropped sharply in price on the international market. Suddenly Rwanda found itself among the many debtor nations required to accept strict fiscal measures imposed by the World Bank and the donor nations. The urban elite saw its comfort threatened, but the rural poor suffered even more. A drought beginning in 1989 reduced harvests in the south and left substantial numbers of people short of food. Habyarimana at first refused to acknowledge the gravity of the food shortage, an attitude that exemplified the readiness of the urban elite to ignore suffering out on the hills.22

The imbalance in wealth and power was a question not just of the usual urban-rural disparities but also of increasingly evident discrimination against Tutsi and against Hutu from areas other than the “blessed region,” that is, the northwest. Habyarimana had established a system of quotas, supposedly to assure equitable distribution of resources and opportunities to all Rwandans. In fact, officials used the system to restrict the access of Tutsi to employment and higher education, and increasingly to discriminate against Hutu from regions other than the north. By the mid-1980s, Habyarimana’s home prefecture of Gisenyi, one of ten in the country at the time, had provided the office holders for one-third of the most important jobs in government as well as virtually all the leaders of the army and security service. Gisenyi and the adjacent prefecture of Ruhengeri enjoyed a similarly disproportionate share of national resources, whether measured in terms of funds for development or places available for higher education.23

Opposition within Rwanda

Confronted by the dramatic economic decline and the evidence of increasing corruption and favoritism on the part of Habyarimana and his inner circle, political leaders, intellectuals, and journalists began demanding reforms. These critics within Rwanda echoed demands for greater democracy being heard elsewhere in Africa and in other parts of the world. They were in turn backed by donor nations that now saw political reform as necessary for economic progress. In July 1990, Habyarimana agreed to discuss change and announced that a national commission would be formed to examine the question. Two months later, a group of thirty-three intellectuals and leaders of the awakening civil society declared that in their view the issue needed no further examination: Rwanda should return to a multi-party system. In that same month of September, four journalists were brought to trial for having published reports of government corruption. They were led by Abbé André Sibomana, editor of Kinyamateka, the oldest and most influential newspaper in the country. In denouncing abuses of power, Sibomana broke with the position of the archbishop and others in the hierarchy, who continued to give Habyarimana apparently unquestioning support.24 After presenting considerable evidence to substantiate their charges, the four were acquitted in a decision that seemed both to confirm the accuracy of the reports and to herald a new era of freedom for the press. The next week, Habyarimana named the members of the commission to examine political reform. Just as these changes were promising greater participation in the political system, the RPF attacked Rwanda.

The RPF Attack

By the late 1980s, the Rwandan community in exile had swelled to approximately 600,000 people,25 most of whom lived in the countries surrounding Rwanda. Except in Tanzania, where the government had encouraged their integration into the local population, the refugees existed precariously, with few rights or guarantees. In Uganda, thousands of refugees had been expelled toRwanda in 1982, only to be pushed back again across the border shortly after. In 1986 Rwandan authorities had declared that the country was too overpopulated to permit the return of the refugees, a statement that helped spark renewed activity in the refugee community. At a meeting in Washington D.C. in 1988 Rwandans affirmed their right to return home, by force if necessary. In 1989 the Rwandan government created a commission to deal with the refugee problem. It met jointly with Ugandan authorities three times, the last in July 1990, and appeared to be making some progress in clearing the way for the refugees to return.

The RPF, however, decided to go home on its own terms, proclaiming its goals to be not just the return of the refugees, but also the ouster of Habyarimana and the establishment of a more democratic government. Its leaders, part of a generation that had grown up in Uganda, were well prepared to launch this effort. Many of them had learned to make war in the forces of the National Resistance Army, where they had helped Yoweri Museveni win control of the Ugandan state. Among them was Paul Kagame, once deputy head of military intelligence for the NRA, who took command of the Rwandan Patriotic Army (RPA),26 the fighting force of the RPF, in the early days of the war. His forces consisted of some seven thousand soldiers, about half of whom were Rwandan refugees who had deserted from the Ugandan army, bringing along their arms and other equipment.27

The Government Response to the Attack

Rumors that the RPF was about to attack had circulated in both Uganda and Rwanda since mid-September 1990. The Rwandan commander at the frontier, aware of these reports, wired headquarters to ask for reinforcements. He got none, leading him and others to speculate that Habyarimana wanted the invasion. On October 1, 1990, the RPF crossed the border, easily overpowered the small detachment there, and headed for the capital.28

The attack offered Habyarimana the opportunity to rebuild his eroding base of power by rallying Rwandans against the enemy. In response to the news, the great majority of people, Tutsi and Hutu opponents of the regime included, came to the support of the government. But Habyarimana understood that the attack posed a risk as well as an opportunity: it might embolden the opposition within the country and even lead to its alliance with the enemy. Rather than rely on a spontaneous coalescing of support from all sides, Habyarimana decided to pursue a more forceful strategy, to sacrifice the Tutsi in hopes of uniting all Hutu behind him.

On October 4, the RPF had advanced a considerable distance into Rwanda but was still forty-five miles from Kigali. That night, however, heavy firing shook the capital for several hours. In the morning the government announced that the city had been attacked by RPF infiltrators who had been driven back by the Rwandan army. Under the pretext of assuring security, the government began making massive arrests in Kigali and elsewhere in the country, eventually imprisoning some 13,000 people. The detainees would be held without charge, thousands of them for months, in deplorable conditions. Many were tortured and dozens died. The last of them were finally liberated in April 1991.29

Many Rwandans and apparently all foreign observers believed the government account of the battle and the infiltration. In fact, the attack had been faked. Habyarimana staged the event to have credible grounds for accusing Tutsi of supporting the enemy. He disclaimed any such intention, declaring on October 5 that there was no question of considering “our brothers and sisters of whatever ethnic group” responsible for what had had happened.30 But certainly he knew and approved of the plan as well as of the arrests that resulted from it. The minister of justice spoke more openly. In the first use in the 1990s of the term that was to become so famous, he declared that the Tutsi were ibyitso, “accomplices” of the invaders. He continued that “to prepare an attack of that scale required trusted people [on the inside]. Rwandans of the same ethnic group offered that possibility better than did others.”31

In accusing the Tutsi, the authorities reverted to the tactics of the 1960s, but in a departure from the earlier practice, they included Hutu as well among the “accomplices.” Unwilling to wait for the scapegoating of the Tutsi to produce solidarity among the Hutu, they sought to hasten the effect by imprisoning Hutu opponents, hoping to silence and perhaps even eliminate some while at the same time intimidating others into rallying to the president.

The faked attack served another purpose: to ensure help from friendly foreign nations. When asked the reason for all the firing on the night of October 4, one Rwandan army officer is reported to have replied, “It was fireworks to welcome our friends, the French,” who did, in fact, arrive that night.32 Pretending that even the capital was at risk, Habyarimana was able to enlist immediate support from Belgium and Zaire as well as from France. The Belgian forces stayed only a month and the Zairian soldiers were sent home for indiscipline, but the French soldiers remained to become a solid support for the Rwandan army and the Habyarimana regime.

With the help of foreign troops, Rwandan soldiers drove the RPF back towards the Ugandan border. As they advanced through the region called Mutara, the Rwandan forces killed between 500 and 1,000 civilians. The unarmed victims were Bahima, a people usually identified with Tutsi, and they were accused of having aided the RPF.33

The government instituted a series of security measures, including requiring citizens to participate in patrols at night and to man barriers to monitor traffic on roads and paths. The neighborhood official, the nyumbakumi, was responsible for enforcing these measures and for keeping track of any strangers who entered his part of the commune. Except in communes adjacent to battle zones, these measures did not last long, but they did help convince people that there was a real danger of enemy infiltrators.

Consolidating the Opposition

The imprisonments of October reinforced the image of the Habyarimana government as a repressive regime and instead of driving Tutsi and Hutu opposition apart, strengthened bonds between them. In a January 1991 letter, prefects urged Habyarimana “to vigorously destroy the manoeuvers of the enemy, both...the INYENZI34 terrorists and those of the opposition that has developed inside the country.” They advised him to “fight openly against what could be called the ‘Kanyarengwe effect’ which poses a serious threat to the necessary solidarity of the BAHUTU.”35 Colonel Kanyarengwe, the important officer who had fled Rwanda in 1980 after accusations that he was plotting against Habyarimana, had joined the RPF and was serving as its president. Because he was a Hutu—and from northern Rwanda besides—his participation in the RPF exemplified the dreaded union of dissatisfied Hutu and the RPF.

Knowing of RPF pressure on the regime, its opponents were encouraged to demand more rapid change. The Rwandan human rights movement was stimulated by the massive arrests at the start of the war. The first of the groups, the Rwandan Association for the Defense of Human Rights (Association Rwandaise pour la Défense des Droits de l’Homme, ARDHO) had been established the night before the RPF attack and faced its first challenge in dealing with the arrests. Two others were founded directly in reaction to the imprisonments: the Rwandan Association for the Defense of Human Rights and Public Liberties (Association Rwandaise pour la Défense des Droits de la Personne et des Libertés Publiques, ADL) developed from a network of those who tried to bring relief to the prisoners and their families and Kanyarwanda was established by former prisoners once they were liberated.36 These organizations quickly began insisting on reforms necessaryto permit full enjoyment of civil and political rights. Donor nations, too, urged Habyarimana to open up the political system, hoping this would speed an end to the war.

In announcing the national commission on reform in July 1990, Habyarimana had anticipated a two-year period of study before it would submit its report. But only eleven months later, in June 1991, he was obliged to accept the constitutional amendment that made multiple political parties legal. Even before the amendment was adopted, opponents began to organize the Democratic Republican Movement (Mouvement Démocratique Républicain, MDR), which would constitute the chief threat to the MRND. Within months another fifteen parties had been formed, the most important of which were the Social Democratic Party (Parti Social Démocrate, PSD), Liberal Party (Parti Libéral, PL) and the Democratic Christian Party (Parti Démocrate Chrétien, PDC).

With the organization of parties, the opposition had structures to mobilize protest against the establishment. Their first goal was to force Habyarimana to accept a coalition government which would give them a chance to share in power. He resisted their demands for some months but after the opposition parties mounted massive street demonstrations early in 1992, he was obliged to begin talks with them. As these negotiations were going on, a group of Hutu announced the establishment of a new party, the Coalition for the Defense of the Republic (Coalition pour la Défense de la République, CDR). They asserted that “no party, no institution, no person had been able to defend the interests of the majority [i.e., the Hutu] publicly and consistently,” and so they must take their fate in their own hands.37 The CDR openly criticized the MRND and even Habyarimana personally for conceding too much to the opposition parties and to the RPF. Despite this criticism, the CDR collaborated frequently with the MRND, leading some observers to conclude that this bitterly anti-Tutsi party existed only to state positions favored by the MRND but too radical for them to support openly.

Habyarimana agreed to incorporate the major opposition parties in a coalition government, which took office in April 1992. In it, Habyarimana continued as president of the republic and the MRND was able to retain nine of the nineteen cabinet posts, including the key ministries of defense and interior. But the largest of the new parties of opposition, the MDR, obtained the post of prime minister as well as two other ministeries. In addition, the PL and the PSD each had three seats and the PDC had one. The new CDR, representing only a small number of adherents, was not included.

Once at the cabinet table, the opposition parties next aimed to divorce the MRND from the state, the natural consequence of introducing a multi-party system. At their insistence, the minister of interior directed administrative officials to show neutrality in the exercise of their functions instead of being cheerleaders for the MRND.38 Once able to count on buildings, vehicles, office equipment, and supplies that belonged to the state, the MRND would henceforth have to provide its own resources. The divorce was faster and more complete in regions where the opposition parties had established a solid base, less so in the northwest where the continued preeminence of the MRND made it futile to protest its privileges. Wherever possible, the MRND naturally delayed yielding its advantage. Radio Rwanda, for example, continued for some time to play MRND songs, supposedly because it had no other tapes in its music collection.

To make their participation in power real and convincing—and hence to draw more adherents to their flags—the opposition parties had to end the MRND monopoly over government posts. They had to deliver to their members the jobs usually associated with controlling the state and they had to be in a position to ensure that the policies they favored would be executed. They quickly put their own people behind the desks in the ministries they headed, but determining appointments in Kigali was not enough. They needed to control at least some of the local administration whose support was usually essential to winning elections. Within a few months of joining the government, the MDR, the PL and the PDC each had gotten one post of prefect. It was even more important for them to have the support of burgomasters, who could do much to sway election results within their communes. This took longer and it was only in February 1993 that the MRND agreed to changing burgomasters in about one third of the communes.

One of the first domains where the opposition ended exclusive MRND control was access to education. In 1991, only 8 percent of Rwandan children were ableto study at secondary school.39 Through the Ministry of Primary and Secondary Education, the MRND had regulated access to government-supported high schools, supposedly assigning places according to quotas for ethnic and regional groups. The quotas were both inaccurately computed and unfairly applied, favoring children from the northwest or those whose families could pay in money or other benefits for access to education. With the April 1992 government, Agathe Uwilingiyimana took office as minister of primary and secondary education.40 A representative of the MDR, she promptly abolished the quota system and decreed that access to higher education would be decided on merit alone. Almost immediately after, she was assaulted by armed men who forced their way into her house and beat her. Thousands of students and mothers turned out to march in support of her new policy.41

In the early months after the parties were established, their supporters saw the new organizations as the hope of the future—for themselves personally as well as for the nation. In a brash and exuberant rush to publicize their cause and to recruit new members, party activists sporting caps and shirts with the party colors held demonstrations and meetings in small commercial centers out on the hills as well as in the capital. Local leaders flew the party flag on poles outside their homes or businesses, proud to be identified as the key persons for mobilizing adherents in that area. Party leaders organized groups of singers or dancers to enliven meetings with musical versions of party propaganda, mirroring the “animation” that had once been the exclusive domain of the MRND.

MRND officials naturally feared the development of opposition parties. The prefect of Butare, for example, wrote his subordinates in early 1992 to warn that parties posed a risk to the “unity of the popular masses.” Like many others at the time, he cast the danger in terms of defeat by the enemy, not in terms of the loss to some rival political party within the country. He insisted that if Hutu opponentscontinued contesting MRND control, the Tutsi would take power.42 MRND leaders at the national level were concerned enough about the threat from other contenders to direct local authorities, still all nominally MRND supporters at that time, to do a poll of political loyalties within some of their districts. In Bwakira commune, sector leaders reported that in some places Habyarimana and the MRND would be chosen by only 50 percent of the voters.43

The MRND authorities did their best to slow the organizing efforts of rivals by using security regulations to hinder their travel and public meetings. They looked the other way when MRND members disrupted demonstrations of the opposition and stole or destroyed their party insignia. In some places they tolerated or even encouraged MRND supporters to assault members of the opposition or to burn and pillage their houses. Seeing the power of the state used for partisan ends, adherents of opposition parties also adopted force as a means of winning the political struggle. Taking political recruits by force or by threat became known as kubohoza or “to help liberate,” an ironic use which suggests that the captive might have been “freed” against his or her will. Originally undertood to mean liberating from the MRND monolith, the term later was used to refer to aggressive action against any political opponent.

The parties organized youth wings which increasingly engaged in violence against rivals. The MDR youth wing, the Inkuba or “Thunder,” led in harassing MRND supporters, sometimes with the help of the Abakombozi, “The Liberators” of the PSD. Confronted with this opposition, the MRND moved to a new level of intimidation by transforming its youth group, the Interahamwe, into a real militia. Besides being more numerous and better organized than the youth of other parties, the Interahamwe received military training from regular soldiers beginning in 1992. They were sometimes backed by the CDR youth group, the Impuzamugambi, “Those With a Single Purpose.” During 1992 and 1993, politically motivatedattacks by Interahamwe and other groups took some 200 lives and injured scores of people in different communities.44

If the target to be “liberated” was sufficiently important, the process could involve rewards as well as threats. In the commune of Nshili, Gikongoro prefecture, for example, an ambitious young teacher named Paul Kadogi decided to join the MDR in part because he was having difficulties with the burgomaster, an MRND stalwart who had held the post for some thirty years. Because Kadogi, described by MRND higher authorities as a “very virulent” propagandist for the MDR, was attracting considerable support among teachers and others in the commune, the MRND dispatched a “mission” in June 1991 to win Kadogi back. The senior member was secretary-general of the Ministry of the Interior and a native of the region. He was assisted by a burgomaster from an adjacent commune who was also a member of the prefectural committee of the MRND and by the sub-prefect of the region. The MRND emisssaries combined what they called “muscular persuasion” with the promise to name Kadogi himself burgomaster if he agreed to rejoin the MRND “with all the people who had followed him into the MDR.” On August 12, 1991 the prefect of Gikongoro “took great pleasure” in writing the minister of the interior to announce that the “recovery” of Kadogi and his numerous followers had been completed. The prefect had just returned from the ceremony installing Kadogi as burgomaster of Nshili where he had “forcefully and enthusiastically” invited all the MDR members in the crowd to follow his example of rejoining the MRND. In his report on the mission, the sub-prefect stressed the effectiveness of visits by important officials from the capital who were native to the region in rallying people to the MRND. The prefect, in his report, assured the minister of the interior that: “We remain vigilant and ready to dismantle in the same way any effort or campaign that might be launched here...by other parties developing at the expense of the MRND.”45

The MDR adherents did not count themselves defeated although it apparently took them some months to recover from Kadogi’s defection. By November 1992, they were ready to use kubohoza and went so far as to attack and take hostage National Policemen. A month later, the police shot and killed a member of the MRD youth group in the same region. This provoked MDR activists in several communes to threaten the sub-prefect and the prefect whom they accused of using the police to destroy their party. The prefect, Laurent Bucyibaruta, protested his complete neutrality and his readiness to permit demonstrations by other political parties, provided their organizers were willing to “take the consequences if another part of the population decides to react against these demonstrations.”46

In this case, the prefect and sub-prefect avoided assault, but other MRND authorities, higher as well as lower in rank, were attacked, particularly in 1992 and early 1993. Several burgomasters were driven from their communes and forced to resign. The minister of youth was assaulted while driving through a commune hostile to him. The home of the minister of labor was attacked in the prefecture of Kibungo.47

The illegitimate use of public powers for private or partisan benefit discredited not just the office-holders, but also the institutions themselves in the eyes of the population. In communes where the burgomaster was accused of governing badly, people refused to pay taxes, the situation in a considerable number of communes by mid-1992. In those places where the land-hungry cultivators had been obliged by the state to cede fields to development projects that brought no visible improvement to their lives, they took back the land by force. In communes where umuganda obligatory work was bringing no benefit to the ordinary people, they began refusing to turn out for the day of labor.

Impunity and Insecurity

When people engaged in kubohoza, they sometimes covered their faces with chalk, wore banana leaves, attacked at the signal of a whistle, marched to a drum and manned barriers along the roads to catch their prey. During the genocide, some assailants did the same things. More important by far than these surface resemblances was the continuation of an attitude spread by kubohoza, an attitude that accepted violence as “normal” in the pursuit of political ends. Just as MRNDofficials frequently tolerated or encouraged violence by MRND members, so did officials of other parties condone or incite the use of force by their supporters. When authorities halted or punished violence, it was often because the perpetrators belonged to political parties to which they themselves were opposed. The National Police and soldiers sometimes refused to assist civilian officials who were attempting to maintain order and sometimes they even launched politically motivated attacks themselves against opponents of the MRND or CDR.48 The judiciary did no better than the executive branch in upholding a state of law. The courts, underfunded and understaffed, rarely functioned as they should have.49

During 1992 and 1993, apparently random attacks by unindentified assailants increased dramatically: grenades thrown into houses, bombs placed in buses or at markets, and mines laid along roads. The Rwandan army general staff issued a press release identifying RPF infiltrators and their “accomplices” as responsible for this violence, an assessment generally accepted by supporters of Habyarimana.50 Those opposed to Habyarimana attributed the attacks to his agents, who, they charged, were operating a death squad which they called by Mfizi’s name of the “Zero Network.” The International Commission of Investigation On Human Rights Violations in Rwanda, a group sponsored by four international human rights organizations that examined the situation in Rwanda in early 1993, concluded that the Zero Network was linked to the highest circles of power in Kigali and was responsible for many of the attacks.51 Whether executed by agents of Habyarimana or by others, the random violence, like the targeted violence of kubohoza, showed Rwandans that the government either could not or would not protect its citizens.52

In the absence of an impartial, effective enforcement of the laws, those who attacked with political motives multiplied their abuses. Common criminals profitedtoo from the laxity of law enforcement to increase assaults and robberies. Firearms had suddenly become easy to get, partly as a result of the war-time increase in the circulation of guns, partly as the result of distribution of weapons by officials. Grenades could be bought at the market for less than U.S.$2.53 The availability of guns and grenades made the work of criminals easier, more certain to be profitable, and more likely to prove fatal for the victims. In some communities, National Police and soldiers raped, pillaged, or even murdered the civilians they were supposed to be protecting.54 Unable to count on protection from the state, law-abiding Rwandans who feared attack because of their politics or their wealth also invested in guns, some of which were registered as required by law, others of which were kept hidden until the genocide.55

After the initial RPF attack in October 1990, the Rwandan government forces, assisted particularly by the French, repulsed the invaders, killing many of them. The RPF regrouped and, in a surprise attack, took the important northwestern town of Ruhengeri in January 1991, but held it for only one day.56 Reduced to only about 3,000 soldiers, the RPF retreated into a series of guerrilla incursions which were met with ripostes from the Rwandan army.57 The combat was punctuated by occasional efforts at cease-fires and negotiations, but it was only after the MDR, the PL, and the PSD joined the government in April 1992 that they were able to oblige Habyarimana to enter into serious negotiations with the RPF. At the sametime, the RPF launched an important offensive in the northeast, apparently to assure a strong position at the start of peace talks. They drove Rwandan army troops back from several communes in Byumba prefecture along with some 350,000 civilians who thus began years of misery as displaced persons. The RPF and the Rwandan government signed a cease-fire at Arusha, Tanzania in July 1992 and in August 1992 they signed the first of a series of agreements that would be known as the Arusha Accords. The Organisation of African Unity (OAU) facilitated the negotiations and agreed to provide a small observer force to monitor the cease-fire.

By the time serious talks with the RPF began in 1992, the Rwandan army had grown to some 30,000 soldiers. An important number of them opposed the negotiations, not just because they did not want to give up the fight, but also because they dreaded demobilization. The thousands of troops who had been recruited since the start of the war had become accustomed to the advantages of the military life. The MRND and the CDR fed their fears by spreading rumors that soldiers would be thrown out onto a disintegrating economy without hope of finding work. The prime minister, Dismas Nsengiyaremye of the MDR, attempted to reassure the troops by talking of using demobilized soldiers in economic development projects, such as draining marshes to obtain new land for cultivation. This proposal incensed the soldiers further; it was just such menial labor that they thought they had left behind in their new military careers.

In May and June, 1992, soldiers mutinied in the northern towns of Gisenyi, Ruhengeri, and Byumba killing scores of civilians and pillaging or destroying hundreds of thousands of dollars worth of property. Soldiers rebelled again briefly in October at the Kanombe military base near the capital.58 Responding to pressure from the military as well as from civilian hard-liners, Habyarimana disavowed the Arusha Accords in a speech in Ruhengeri on November 15. Making clear that he did not intend to implement the agreement that he had signed three months before, Habyarimana called the Accords “a scrap of paper.”

In principle prohibited by law from membership in political parties, soldiers and police nonetheless did not hesitate to demonstrate their political leanings. Habyarimana himself was only the most obvious case, serving until 1992 as general and commander-in-chief of the Armed Forces while also being president of the MRND. Particularly those soldiers who shared a northern origin with Habyarimana, of whom there were many, put loyalty to the president above all else. Some officers of the army general staff promoted fear and hatred of Tutsi and ofHutu opposed to Habyarimana both among the troops and among the civilian population. In early December 1991, the high command of the Rwandan army issued two press releases that proclaimed in a pro forma way their support for democratization and neutrality towards all political parties. But the military leaders then went on to condemn Rwandans who “knowingly or unknowingly, aided the enemy under the cover of political party activities.” They declared that newpapers critical of the president were subsidized by the RPF. They blamed RPF infiltrators and their “acolytes” for the increase in crime and acts of random violence and they concluded one press release by asking the secret police to “neutralize all collaborators identified with the enemy.”59 Col. Léonidas Rusatira, then secretary-general of the Ministry of Defense, apparently opposed the broadcast of these releases, but he was overruled by Habyarimana himself who decided to make them public.60 The minister of the interior circulated the first of these press releases, directing that burgomasters make its contents widely known. The prefect of Kibuye, passing on the order, told burgomasters to “use it [the press release] to its full value in meetings to raise the consciousness of the population about the ideals of peace and unity.”61 The release must certainly have had the opposite effect, itself fueling the “ethnic and regional tensions” that it accused opponents of fostering.

On September 21, 1992, Colonel Déogratias Nsabimana, chief of staff, sent a top secret memorandum to his commanders identifying and defining “the enemy.”62 The memorandum was part of a report from a commission of ten officers established the previous December to examine how to defeat the enemy “in the military, media and political domains.” Among the measures recommended by the commission was the removal of some high-ranking officers who held these posts by virtue of their connections to members of the akazu, particularly Madame Habyarimana, rather than by virtue of their military abilities. Habyarimana had accepted their recommendations in June 1992 and had obliged a number of officers to retire, among them Colonels Serubuga and Rwagafilita.63 The memorandumremained restricted to a small circle of high-ranking officers until Nsabimana ordered its dissemination in September, several weeks after the signing of the first of the Arusha Accords. Rwandan military authorities at this time feared a new RPF offensive was being prepared and Nsabimana hoped the memorandum would “lead our men to be more vigilant and to not count on political negotiations alone.” He ordered:

You will distribute this document widely, insisting especially on the sections relating to the definition of the enemy, identification of the enemy, as well as the groups within which the enemy is recruited. You will inform me of the impact made by the contents of this document on the men under your orders.

The report divided the enemy into two categories, the principal enemy and partisans of the enemy. The principal enemy was:

the Tutsi inside or outside the country, extremist and nostalgic for power, who have NEVER recognized and will NEVER recognize the realities of the 1959 social revolution and who wish to reconquer power by all means necessary, including arms.

The partisans of the enemy were defined as anyone who supported the principal enemy. Like the December 1991 press releases, the document made the necessary nod towards democratic openness:

Political opponents who want power or the peaceful and democratic change of the current political regime of Rwanda are NOT to be confused with the ENI [enemy] or with partisans of the ENI.

Again like the earlier communiques—and sometimes in the same language—the fourteen page document then went on to condemn Tutsi and those Hutu who opposed Habyarimana and his party. Nowhere did it caution against confusing the RPF as a political group with Tutsi as an ethnic group. In several places, it used “Tutsi” as equivalent to enemy. As one of the advantages of the enemy, it listed “A single political will and a single political ideology, which is Tutsi hegemony.”

The document deplored the loss of Hutu solidarity, which it blamed on enemy machinations rather than on understandable resentment of the corruption and repression of the Habyarimana regime. It listed the establishment of multiple political parties as an advantage for the enemy and warned that infiltrators had convinced these parties to support the RPF. Repeating the accusation of theDecember 1991 press release that the enemy was sharpening conflict between individuals and regions, the memorandum asserted that opponents were “turning public opinion from the ethnic problem to the socio-economic problem between the rich and the poor.” It stated that the enemy and its partisans were recruited primarily among:

· Tutsi refugees

· the NRA (Ugandan army)

· Tutsi inside the country

· Hutu dissatisifed with the regime in power

· Unemployed people inside and outside the country

· Foreigners married to Tutsi wives

· the Nilo-Hamitic people of the region

· criminals in flight [from the law]

The document warned that the enemy had infiltrated the government and had corrupted various officials by offering them advantageous business deals, which was easy for them to do because the enemy predominated in business circles. It identified a number of “enemies” by name, including Evariste Sissi and Antoine Sebera.64

Many of the themes of this document sent to the soldiers on September 21 are echoed in a CDR tract issued the next day. In its “Notice No. 5,” the CDR warned of the dangers from enemies inside Rwanda, who were supposedly aiding the RPF. It asserted that these enemies had highly placed friends in the government, who were permitting them to work against the interests of the great majority, the rubanda nyamwinshi. Among the enemies named are the same Evariste Sissi and Antoine Sebera who were cited in the military document. The CDR finished by demanding action:

The CDR party calls upon the government and the president to deal with this problem. If it does not, the great mass [rubanda nyamwinshi] cannot stand by and do nothing. An enemy is an enemy. Anyone who cooperates with the enemy is a traitor to Rwanda.65

The similarities in the statements of CDR radicals and of high military authorities foreshadowed their later cooperation which made the genocide possible.

64 Col. Déogratias Nsabimana to Liste A, September 21, 1992.

65 Itangazo no. 05 ry’ishyaka CDR, 22 September 1992 (International Commission).