Flexibility in Domestic Organization and Seasonal Migration among the Fulani of Northern Burkina Faso
Hampshire, Kate, Africa
Migration patterns among the Fulani of Burkina Faso have changed over recent decades from predominant transhumance, involving whole families, to seasonal rural-to-urban labour migration of young men. This article uses a combination of quantitative and qualitative data to examine the relationships between the new forms of migration and domestic organization. Specifically, it asks the following questions: (1) How do households accommodate the temporary loss of productive members? (2) Does the out-migration lead to new forms of domestic organization, and to changing roles and power relations within sending households? Various forms of flexibility in domestic organization are identified, which serve to maintain viable economic units in the face of the temporary absence of substantial numbers of young men. These include: flexibility in the processes of household division; rapid, temporary restructuring of domestic units; and drawing on extra-household support networks. One consequence of this flexibility is that intra-household gender divisions of labour and power have remained largely unchanged in the face of seasonal labour migration. The extent to which this will remain the case if migration becomes more widespread is uncertain.
Les schemas de migration observes chez les Foulanis du Burkina Faso ont evolue au cours de ces dernieres decennies, la transhumance autrefois predominante, impliquant des families entieres, faisant place a une migration rurale-urbaine saisonniere de main-d'oeuvre masculine jeune. Cet article se sert de donnees quantitatives et qualitatives pour examiner les relations entre les nouvelles formes de migration et d'organisation domestique. Il pose en particulier les questions suivantes : (1) Comment les menages s'accommodentils de la perte temporaire de membres productifs ? (2) Cette emigration externe conduit-elle a de nouvelles formes d'organisation domestique et a une evolution des roles et des rapports de force au sein des menages de migrants ? L'article identifie diverses formes de flexibilite dans l'organisation domestique destinies au maintien d'unites economiques viables en l'absence temporaire d'un nombre significatif d'hommes jeunes. Parmi ces formes figurent : la flexibilite dans les processus de division des taches menageres; la restructuration rapide et temporaire des unites domestiques; le recours aux reseaux de soutien extramenager. L'une des consequences de cette flexibilite est le caractere largement inchange de la division intramenagere des taches et du pouvoir entre les sexes face a la migration de main-d'oeuvre saisonniere. On ne peut en revanche pas savoir s'il en restera de meme, et dans quelle mesure, si la migration prend de l'ampleur.
There is a long tradition of rural-urban migration in much of sub-Saharan Africa (Caldwell 1969; Murray 1981; Cordell et al. 1996). The impact of rural out-migration on livelihoods, and on the development of sending areas, has been an area of much debate. Many studies suggest that the exodus of young, productive adults leads to the underdevelopment of rural areas and, sometimes, to the collapse of livelihoods (Maliki et al. 1984; Cleveland 1991; Cordell et al. 1996). Others paint a more optimistic picture, in which rural production is boosted by injections of capital from migrants' earnings (Lucas 1987; Ruthven and Kone 1995; Hampshire and Randall 1999). Rural outmigration may also have impacts on social and domestic organization in sending areas. How do households accommodate the loss, temporary or permanent, of some of their most productive members? Does rural outmigration lead to new forms of domestic organization, and to changing roles and power relations within sending households?
Several studies have shown that larger households are in the best position to accommodate out-migration, since labour can be spared more easily (Saint-Pierre et al. 1984; Gaye GuinGuido 1992; Toulmin 1992; David 1995; Knerr 1998). The demographic composition of the household is also important. Findley (1994) round the number of young men in the household, rather than overall household size, tobe the critical factor in Mali, where it is predominantly young men that migrate. Working among the Wodaabe Fulani in Niger, Maliki et al. (1984) suggest that the normal processes of household division may even be delayed in order to facilitate out-migration, following a crisis such as severe drought. David and Niang (1995) also found evidence of out-migration of men affecting household structure in Senegal, with normally separate consumption and production units joining together in the dry season, during the absence of the compound head. Other studies have pointed to changes in the opposite direction. Murray (1981) suggests that in Lesotho, a country that has long experienced extremely high levels of labour migration, migration has led to domestic fragmentation. Francis (1995; 2002) found that, in western Kenya, out-migration of men was accompanied by increasing importance of the conjugal unit relative to the extended household.
If out-migration is accompanied by changes to the size and structure of domestic units, dais might affect intra-household divisions of labour. Okoth-Ogendo (1989) claims that a common consequence of male out-migration is to shift the agricultural workload towards women, thus increasing their overall burden. A similar observation was made by Monimart (1989) in Mauritania, who found that out-migration of men added substantially to women's labour burden. Very few of the benefits from migration filtered through to women who, to make up for the loss of men's productive labour, were obliged to engage increasingly in off-farm activities. Gisbert et al. (1994) found that women bore the brunt of men's out-migration in Bolivia with an increased agricultural workload. David (1995) contend that no such generalizations can be made. In Senegal, they found that out-migration of men increased women's workload, while among Mossi groups in Burkina Faso, it was male relatives of the migrating men, rather than their wives, who covered the labour deficits. Both David and Gisbert et al. suggest that the long-term security of rural livelihoods might be put at risk because non-essential maintenance tasks tend to be neglected by remaining household members under excessive work pressure.
Power relations within households might also change. Where out-migration of men leads to the creation of de facto female-headed households, there is potential for women's control over resources and decisions within households to increase. Such processes have been described in southern Africa (Timaeus and Graham 1989) and in southern Nigeria (Mabogunje 1989). However, David's multiple-country study (1995) shows that the creation of female-headed households does not automatically lead to changing power relations. In the Sudan, while migrants' wives made small everyday decisions over farm management, more important decisions were postponed until the men's return. In Mali, wives were usually incorporated into their husbands' extended families and were subject to decisions made by his kin. Senegalese women took over their husbands' workload, but were not empowered to make even minor decisions without seeking authorization from their husbands. Dependence on remittances from migrants can also shift intra-household power balances. Studies in southern and eastern Africa (Sharp and Spiegel 1990; Francis 1995; Francis 2002) have shown that male migrants usually have exclusive control over their earnings. When coupled with declining productivity in domestic agricultural production, this can lead to an erosion of women's spheres of authority and responsibility within the household.
This article explores the relationships between out-migration of men and domestic organization among the Fulani of northern Burkina Faso. Unlike most of the other cases discussed above, rural-urban migration is a very recent phenomenon for Fulani in this area. Before the 1970s, the main form of migration was transhumance, usually involving whole families. During the 1970s and 1980s, many Fulani in the west African Sahel suffered serious loss of agro-pastoral livelihoods, partly as a result of major droughts, and partly due to important socio-political changes in the newly emerging independent state of Burkina Faso. Since that time, the importance of transhumance has been declining. Instead, men have begun migrating to cities on a seasonal basis to look for work. Over the past three decades, seasonal economic migration has become more widespread, and is of increasing importance to people's livelihoods (Hampshire and Randall 1999; Hampshire 2002).
These rapid and recent changes in migration patterns offer an excellent opportunity to examine the relationships between migration and domestic organization, since the changes and their consequences are well within living memory for most adults. The study also offers an interesting contrast to those of the longer-term, longer-established migrations typical of southern and eastern Africa, which have received much more attention from researchers. A combination of statistical analysis with detailed qualitative data from interviews is used to offer new insights into the relationships between migration and domestic organization in an area of rapid social and economic change. In particular, the following research questions are addressed:
1. What are the processes of household division and formation among the Fulani of northern Burkina Faso? Which factors influence these processes and how flexible are they?
2. What is the relationship between domestic organization and new forms of migration among the Fulani? In particular, how do households accommodate the temporary loss of young men who leave the Sahel for labour migration during the dry season?
3. Does this temporary out-migration of young men lead to new forms of domestic organization, and to changing gender roles and power relations within domestic arenas?
STUDY AREA AND METHODS
Northern Burkina Faso has a typically Sahelian climate and ecology. It is a semi-arid area, with a short rainy season from late June to early September (Barral 1977; Claude et al. 1991). The Fulani constitute about a quarter of the population of the Sahel region of Burkina Faso (INSD 1991). There are two major 'classes' or status groups of Fulani: the Fulbe, a high status group associated traditionally with pastoralism, and their erstwhile slaves or serfs, the Riimaaibe. There are also smaller numbers of marabouts and various caste groups: blacksmiths, potters and bards. In the Burkinabe Sahel there are three major lineages/clans (lenyi) of Fulani: Djelgobe, Gaobe and Liptaako, the names deriving from their perceived place of origin. The main economic activities of the rural Fulani in this area are extensive pastoralism and rain-fed agriculture during the rainy season.
The data presented in this article come from 18 months of intensive fieldwork carried out by the author in 1994-6. Following a pilot visit to the area in 1994, a quantitative single round demographic survey (SRDS) was conducted in April-June 1995, in 40 villages or camps, on a sample of 8834 individuals in 834 ngure or baadeji ('households'--see below). Its purpose was to establish baseline data on fertility, mortality, migration and domestic organization. Data on seasonal/temporary migration were obtained by asking household heads or (where heads were absent) other household members about any household members who were absent at the time of the survey, or who had been away on temporary migration in the preceding year. Information on longer-term and permanent out-migration (including migration of whole households) was obtained using siblinghood methods of estimation (Zaba 1985), in which adults are asked about brothers/sisters who are living (temporarily or permanently) outside of the Burkinabe Sahel. The survey was conducted in Fulfulde using three trained enumerators.
A follow-up multiple round study (MRS), conducted between November 1995 and December 1996, took a structured sub-sample of the survey population. Six villages were purposively selected, two from each major lenyol (Djelgobe, Gaobe and Liptaako), chosen to cover the full range of spatial mobility patterns. The Djelgobe and Gaobe villages contained only Fulbe (there are very few Rimaaibe Goabe or Djelgobe in this area), while the Liptaako villages contained both Fulbe and Rimaaife. In subsequent analysis, these are presented as four different lenyi or ethnic sub-groups: Fulbe Djelgobe, Fulbe Gaobe, Fulbe Liptaako and Rimaaife Liptaako.
The sample comprised 1,224 individuals (117 'households'). Each village was visited for one week every two months. A multiple-method approach was used, including:
* detailed life history interviews, including migration histories, with all men (N = 329) and women (N = 307) aged 15+;
* follow-up interviews with adult men and women, every two months, to track changes in migration and social/domestic organization, plus reasons for, and interpretations of, any changes;
* household history interviews with all babaade'en (household heads/joint heads) (N = 135);
* wide-ranging, informal group discussions, covering migration, social change and social and domestic organization (among other topics);
* participant observation (I stayed with families in each of the six villages during fieldwork), keeping detailed field notes.
All interviews were conducted in Fulfulde by the author, in most cases with the help of a French-speaking interpreter.
CHANGING FORMS OF MIGRATION AMONG THE BURKINABE FULANI
The Fulani are renowned as a highly mobile group, strongly associated with transhumance, moving long distances with cattle in search of pasture and water (Gallais 1975; Barral 1977; Bernus 1988, 1991; Milleville 1991; Raynault 1997). Indeed, as Riesman (1984) indicates, mobility of various kinds has been the major strategy used by Fulani to cope with a variable and unpredictable natural environment. However, major herd losses, associated with the droughts of 1973 and 1984, have led to substantial changes in this respect. In the SRDS, 29.4 per cent of households had at least one person going on transhumance in the preceding year, with pastoralist households not surprisingly more likely to send someone (75 per cent, N = 152) than those combining pastoralism with agriculture (23.3 per cent, N = 520) or those involved predominantly with agriculture (1.6 per cent, N = 126). However, migration histories indicate that numbers of people going on transhumance are substantially fewer than before the droughts. Other characteristics of transhumance have changed. The distances moved are much shorter, often less than 10 km in a year, with the main purpose being to avoid fields under cultivation; and movements are typically confined to the rainy season, rather than all year round. There has also been something of a shift away from whole families going on transhumance towards one or two young men going alone with the cattle.
However, over recent decades, there has been an increase in other kinds of migration undertaken by the Burkinabe Fulani. Many of these (but not all) are associated with diversification of livelihood strategies (cf. Bolwig and Paarup-Laursen 1999). Table 1 shows, for 1996, the proportions of Fulani men and women of different age groups participating in different kinds of temporary migration, not directly associated with agro-pastoral production.
With fewer animals to tend, the dry season has become a slack season for many, and a good window of opportunity to seek supplementary sources of livelihood. Today, many Fulani men leave each year following the harvest to work in cities, returning to sow the following year's crop at the beginning of the rainy season. Such migration is known locally as exode. (Exode is a French term, used also by the Fulani, to mean temporary migration to cities. The Fulfulde term for migration, egugol, is far less specific in its meaning, so the French term is used in preference here.) For the purposes of the SRDS, exode is defined as a migration beyond the Sahel region of Burkina Faso, for between one month and two years, with the intention of earning money. In 1995, 15.8 per cent of the 2,149 men aged 18-64 surveyed had been away en exode during the 12 months preceding the survey, and 36.6 per cent of men in this age group had gone at some point in their lives (Hampshire and Randall 1999). The vast majority of migrants (79.5 per cent in 1994-5) go to Abidian in Cote d'Ivoire, with smaller numbers visiting major cities in Burkina Faso and elsewhere. In addition, there are a few men who undertake rural-rural seasonal migration: these are mainly Djelgobe herders who practise contract herding in northern Cote d'Ivoire (cf. Bassett 1988; Hagberg 2000), but their numbers are very few compared with those going to cities. Almost all migrants are away during the dry season and return to their villages for the rainy season, the median length of exode being five months. The other major form of subsistencerelated migration is young men going to work in local gold mines; this was also noted by Bolwig and Paarup-Laursen (1999) in their study of Riimaaibe in a village near Dori.
In addition to these temporary movements, some whole families of Fulani have left the Sahel region of Burkina Faso permanently, usually to continue pastoralism/agro-pastoralism elsewhere. In SRDS data on permanent out-migration, or any out-migration lasting more than two years, 6.7 per cent of men and women reported having at least one brother who had moved permanently out of the Burkinabe Sahel. This was more common among pastoralists (10.1 per cent, N = 703) than among agro-pastoralists (6.5 per cent, N = 3301) or cultivators (4.7 per cent, N = 793). Most commonly, whole families left the region in the droughts of 1973 and 1984, never returning, although smaller numbers have left at other times, many to become contract herders in the south of Burkina Faso (see Bassett 1994; Hagberg 2000). The numbers of individuals/families migrating for more than two years (that is, for a longer period than exode) but not permanently were extremely small: extended migration was usually associated with a failure of pastoral livelihoods, and a sense of shame (semteende), which made it difficult for people to return.
In this article, I focus on exode, rather than other types of temporary migration, because: (1) it involves much greater numbers of people (mainly young men); and (2) it removes these young men from the Sahel, and thus from domestic production for several months of the year. This is in contrast to gold mining: the gold mines of Essakan are relatively nearby, some 40-50 km away from most of the study villages, and so young men move back and forth between their homes and the mines, typically spending a week or two at the mine, followed by some time back at the homestead.
It is beyond the scope of this article to present a detailed discussion of the factors influencing people's decisions to go en exode, and what exode means to the Fulani of this area; this information is available elsewhere (Hampshire and Randall 1999, 2004; Hampshire 2002). In brief, while many young Fulani men go en exode each year, there are big differences in participation in exode among the Fulani, summarized in Table 2. First, the subsistence system makes a difference. There are no 'pure agriculturalists' in the population studied (even among the Riimaaibe), as everyone has at least some sheep or goats. However, those involved primarily in agricultural production are much more likely to leave than are pastoralists or agro-pastoralists. This is explained by the different labour constraints of agriculture and pastoralism: rain-fed agriculture is only possible during the short rainy season, leaving a large window of opportunity during the dry season to leave. Pastoralism, on the other hand, is a year-round activity, with labour demands peaking in the hot, dry season, when water and pasture become scarce (White 1991).
Second, there are important differences in exode between the various lenyi of Fulani in the area. In particular, Fulbe Djelgobe are far less likely to go en exode than any of the other groups. This arises because of the different ways in which the semteende (shame) associated with not being able to subsist from pastoralism (or at least agro-pastoralism) alone is interpreted within the different lenyi of Fulbe. Pulaaku is a concept that describes the sense of Fulbe identity. For any Pullo in this area, part of what it means to be a Pullo is to own cattle and make a living through pastoralism (see also Stenning 1959; Riesman 1977; De Bruijn and Van Dijk 1995). The reality, though, is that most cannot achieve this, since few families have a big enough herd for this to be their only, or even major, source of livelihood. For the Gaobe and Liptaako Fulbe, exode, while attracting some degree of semteende, in that it entails engaging in work other than herding/pastoralism, is regarded as a necessity. The expressed hope among most Fulbe Gaobe and Liptaako migrants is that exode will be a temporary measure, in order to acquire the prosperity needed to return to pastoralism. In fact, this hope is rarely fulfilled, for exode usually produces little by way of returns through remittances--in most cases the main economic benefit to the sending household is in relieving the pressure on the granary over the dry season. Nonetheless, this myth of an ultimate return to the pastoralist ideal sustains a high number of Fulbe migrants from the Liptaako and Gaobe clans, who are as likely to migrate as young Riimaaibe men, for whom there is little or no stigma attached to exode. The Fulbe Djelgobe, by contrast, attach much more value to the practice of herding, and are locally renowned as the most expert of herdsmen. For a Pullo Djelgojo, to leave en exode to work in a city, even for the briefest of periods, attracts so much semteende as to make it almost impossible to consider while other options exist. Fulbe Djelgobe without sufficient cattle of their own usually try to take on contract herding, locally if possible. Others have migrated permanently (to southern Burkina Faso, for example) to do contract herding elsewhere.
Finally, Table 2 shows that household size is a critical factor in determining whether or not an individual goes en exode. Men from larger households are much more likely to leave than those from small households. Indeed, both total household size and the number of adult men in the household are independently significant predictors of individual migration. This pattern is repeated elsewhere. Several authors have noted that larger households are in a better position to accommodate out-migration of men than smaller ones, for whom loss of productive labour may critically undermine domestic subsistence (Saint-Pierre et al. 1984; Gaye GuinGuido 1992; Findley 1994; David 1995; Knerr 1998). The relationship between household formation, household size and exode is central to this article, and is explored in detail below.
There has thus been a substantial shift in the migration behaviour of Fulani agro-pastoralists over the past thirty years, from transhumance movements of whole families to seasonal labour migration of young men, with other members of the family being left behind. This pattern is typical of broader changes in migration streams across Africa, identified by Okoth-Ogendo (1989) as becoming increasingly individual-centred and tied into the wage economy. What impacts have these recent changes had on social and domestic organization among the Fulani? How have Fulani households responded to the temporary loss of substantial numbers of young men each year?
FULANI DOMESTIC ORGANIZATION AND PROCESSES OF HOUSEHOLD DIVISION/FORMATION
Households are notoriously difficult to define, even in relatively settled populations. The basic principles are more or less agreed upon: a locus of production, distribution, transmission, reproduction and co-residence (Wilk and Netting 1984; Roberts 1991). However, among the Fulani, as with many other populations (see, for example, Harpending and Pennington 1990), these units do not necessarily coincide, and there is no single Fulfulde term which translates as 'household'. Moreover, African households are not isolated entities, but operate within wider politico-jural units such as lineages (Fortes 1958; Guyer 1981; Roberts 1991). Fulani society is organized essentially on a system of patrilineages (for details in other Fulani populations, see Hopen 1958; Stenning 1958, 1959; Riesman 1977; Bonfiglioli 1988) although, as some commentators have pointed out, there are certain features of social organization that do not conform to strictly patrilineal forms (Dupire 1963; de Bruijn 1997).
The major unit of domestic organization for the Burkinabe Fulani is the wuro (plural ngure) among the Djelgobe and Gaobe or baade (plural baadeji), among the Liptaako (both Fulbe and Riimaaibe). The two terms are more or less equivalent, and refer to the largest unit of co-residence and regular cooperation in herding, agriculture and consumption. A wuro/baade is typically constituted of a group of agnatically related men, with their wives and children, extending over two or more generations. (1) Baade refers specifically to a courtyard or the collection of huts, and their inhabitants, found within the walls of a courtyard.
The head of a wuro or baade is known as a babaade (plural babaade'en). The literal meaning of this word is 'father of the baade' (baaba baade), although it is used equally to mean 'father of the wuro' among the Fulbe Djelgobe and Gaobe. A babaade is a married man, who has control over a herding or agricultural unit. Within a wuro or baade, there may be some division of herding and agricultural units, in which case there will usually be more than one babaade, although all the babaade'en within a baade/wuro (brothers, fathers and uncles) will usually cooperate closely in productive tasks. Other ngure/baadeji have a single babaade, and all herding and agricultural work operates as a single unit.
A wuro or baade may contain several smaller cooking units (fayannde) and huts or tents (cuudi, singular suudu), even if herds and fields are not divided. In the case of a wuro/baade with a number of married brothers, the wives might form several different fayannde. As De Bruijn (1997) notes, fayannde are organized around, and by, women, as are cuudi. In the case of Djelgobe and Gaobe, who live mostly in tents, a woman brings her suudu with her on marriage and takes it away with her again if she divorces.
Most earlier ethnographies of Fulani groups, particularly in Niger, describe a clear normative framework for household formation (Stenning 1958, 1959; Hopen 1958). Stenning, for example, identifies a number of key stages for the domestic cycle of the Wodaabe, emphasizing the importance of the parallel evolution of people and animals. According to Stenning's account, the process of herd division (and thus the establishment of a new household) begins when a new wife returns to her husband's family after the birth of their first child, and is complete after the birth of enough children to provide sufficient labour for the new household to be economically viable. This has been modified to some extent by more recent accounts, which describe a similar ideology, but a more fluid reality in which household formation is subject to external economic and political circumstances (Maliki et al. 1984; Bonfiglioli 1988).
Among the Burkinabe Fulani, there is no fixed point at which a man becomes a babaade. The processes of domestic division and formation are set within a framework of flexibility and ambiguity, even at a normative level, and the point at which herds, fields and granaries are divided is variable. Most people agree that complete separation should not occur before the death of the father although, in some cases, sons may gain a considerable degree of autonomy before this. Following the father's death, brothers may choose to separate or to stay together in a single wuro/baade, with varying degrees of economic cooperation or separation. The following interview extracts demonstrate the diversity of opinions about when a household should divide:
A: When a man is old and his sons are married, he should divide his goods between his sons, otherwise he may die without his wishes known (Old Gaojo man).
B: While the father lives he is responsible for the important decisions even though his sons, if married, may begin to take some responsibility too. When the father dies the brothers may split or they may stay together: nothing is fixed. It is good to stay together, but sometimes the brothers, or more often their wives, argue, so they divide. Brothers will divide if one is lazy and the others begin to resent him freeloading (Young Gaojo man).
C: People do not have children so that they will leave them in their old age (Old Djelgojo man).
Ambiguity and variability is also demonstrated in examining the actual practice of household division. In cases A and B below, the sons were married, but household division had not begun until the father's death. In case C, by contrast, the beginnings of household division are evident immediately following the marriage of the oldest son, while the father is still alive and well. The implication is that there is considerable flexibility and room for manoeuvre in household division and formation in this population.
Case A: A Liptaako blacksmith's baade
Until late 1995, the baade consisted of the elderly blacksmith, his wife and their children, plus his oldest son from a previous marriage, who was himself married with two young children. The old man died at the end of 1995 and, during the following year, the tension increased between the two halves of the household: the widow and her children, and the oldest son and his family. Eventually, following a major dispute, the baade split and the two halves went their separate ways.
Case B: A Riimaaibe Liptaako baade
This case concerns a very large baade: four young married brothers, the oldest of whom is babaade. Each of the brothers has one or two children. Their father died at the end of 1995, leaving two widows, neither of whom is the mother of the four brothers. Since the death of the father the household is in a great state of flux. The wives of the four brothers do not get on well and have begun cooking separately for their husbands and children, rather than each taking it in turn to cook for the whole baade. This they did of their own accord, without consulting their husbands. The husbands claim this is beyond their control and that, if it continues, it will soon become difficult to keep a common granary as arguments will arise as to how much each wife is taking. It looks likely that a total split will come before long.
Case C: A Fulbe Gaobe wuro
The wuro in question consists of the head, an old man in his sixties, his two wives, and their several sons, the oldest of whom are approaching adulthood. Until 1995 the whole wuro operated as a single unit (herding, cultivating and cooking together), with the old man as babaade. There was a single granary where all the millet was kept and the two co-wives took it in turns to cook, each taking millet from the common granary. In 1995, the oldest son married, and decided he was no longer prepared to cultivate millet for his mother's co-wife. He and his brothers built their own granary into which they put millet they had grown for their mother's private use only. The sons of the other wife followed suit, leaving the old man to cultivate alone. At the time, the household herd was still together, but a more complete split seemed imminent.
Quantitative analysis substantiates these insights into the variable process of division of domestic units. Marriage is clearly a precondition for becoming a babaade: of the 90 never-married men (aged 18+) in the MRS sample, only one was a babaade (the circumstances were exceptional): these 90 unmarried men are excluded from further analyses. Table 3 confirms that a man is most likely to become a babaade if his father is dead and if he has no living older brothers. This is hardly surprising given the above discussion. However, more important to note is that, while certain demographic facts (age, status of father/older brothers) make babaade-ship more or less likely, nothing is immutable, and there remains considerable flexibility in the process of household division. Within each category of men in Table 3, there is variation in babaade status.
What other factors, then, influence whether, when and how a household divides? Household division, and the reasons for it, are not always easy to discuss: initial questions regarding who herds with, or hoes with, whom are often met with: 'miin foo gooto' ('we are all one'). Decisions to split a household or some of its functions, are often taken implicitly rather than explicitly, and come about through actions. When a babaade becomes old, his sons may begin to take increasing amounts of responsibility for domestic production. Over time, it becomes understood that they have become babaade'en in their own right. Or, as in case C above, one brother may simply decide to build his own granary, which then forces a split.
Decisions to divide a baade/wuro or for a man to become a babaade are theoretically for men to take. In many cases, however, women wield considerable influence by simply refusing to cooperate with each other, thus forcing a split. Case B above is typical: if wives of different brothers decide to cook separately, rather than together, this makes it increasingly difficult for brothers to continue to share a granary, and often ultimately results in the division of field and herds.
In people's own accounts of decisions (explicit or implicit) about when or whether to divide the household, individual factors of personality and temperament were most prominent. Accusations of laziness and arguments (often between brothers' wives) emerge as the most frequent explanations. However, there are also some differences in household formation and division based on lenyol/status group, household subsistence system and, especially, on wealth (see Tables 4-6).
Riimaaibe Liptaako have significantly larger households than any of the Fulbe groups in the study population (Table 4). The tendency for larger households among Riimaaibe as compared to Fulbe populations has been documented by other authors (Bolwig and Paarup-Laursen 1999, for example). However, Riimaaibe men are no less likely to be babaade'en than the Fulbe groups (Table 5); instead, there is a pattern of Riimaaibe having larger baadeji with several babaade'en. The interviews conducted showed no major differences in the general processes of household formation between the different lenyol/caste groups, other than that Riimaaibe babaade'en are rather more likely to stay together within the same baade, even after separation of herds/fields, than is the case among the Fulbe. Pastoralist ngure/baadeji are, on average, slightly smaller than those practising agro-pastoralism or agriculture (Table 4), but a smaller proportion of pastoralist men are babaade'en than among agriculturalists (Table 5). When the factors known to affect babaade status (discussed above: marital status, age, status of father and older brothers) are controlled for, there are no statistically significant differences in proportions of men that are babaade'en based on lenyol/caste or subsistence strategy differences (Table 6).
By contrast, wealth does exert an important effect on household division and formation. Defining and measuring wealth among the Fulani is a complex process, which is discussed in detail elsewhere (Hampshire and Randall 1999): briefly, the measures used here are based on within-village wealth ranking by key informants, with livestock holdings as a key component. Wealthier households are substantially and significantly bigger than poorer ones (Table 4). Commensurately, men from wealthier households are less likely to become babaade'en than those from poorer households (Table 5), and this relationship holds when controlling for other relevant independent variables (Table 6). The reasons for this became clear from interview data. A household with a large herd and/or successful agriculture provides a strong magnet, such that dividing and going off alone is a less attractive prospect than it might be for a young man from a very poor household, for whom it might make sense to try to go it alone. As one young Djelgojo man put it, 'When the cows are all gone, what is there to keep me here [i.e. in the same wuro]?'
The most important point arising from this section is that household division and formation among the Fulani of Burkina Faso is a flexible and variable process. While certain factors (death of the father, lack of older brothers, poverty) make household division more likely (either creating a new baade/wuro or having a new babaade within an existing baade/wuro), nothing is immutable. In every case, household division requires individuals to make decisions (implicitly or explicitly), rather than simply following a set of prescribed rules.
RELATIONSHIPS BETWEEN HOUSEHOLD FORMATION AND EXODE
From the discussion above, two points become clear: first, that exode is associated with large households (gure/baadeji), and, second, that, subject to some constraints, processes of household division and formation are very flexible and variable. In this section, I explore the relationship between exode and household formation. Table 2 makes it clear that men from larger households are more likely to go en exode than those from smaller households. Do men simply take advantage of being in large households to go en exode, or might exode itself be a reason to delay household separation? Is maintaining a large household ever used as a strategy to facilitate exode? The same problem can be stated from an individual perspective. Are men who, for some other reason, have not yet formed a separate household taking advantage of their position in order to go en exode? Or are some men delaying household division in order to allow them the freedom to go en exode?
Qualitative data suggest that both of these are true. Some variations in household size are due simply to social and demographic processes unrelated to migration. Households tend to vary in size according to the stage of the development cycle (Fortes 1958). Moreover, simple demographic differences, such as numbers of brothers or uncles, play a role. It is easier to create or maintain a large baade/wuro if one has several brothers and other close male kin.
Oral histories indicate that variability and delay in domestic division and formation are nothing new. Even before the social and economic upheavals of the 1970s and early 1980s, there was apparently considerable variation in when a married man began to take economic responsibility for his own wuro/baade. In this respect, the situation in Burkina Faso is different from that described by Bonfiglioli (1988) and others in Niger, where there had been a much clearer and more prescriptive normative code concerning when a man should begin to form his own wuro, although this code could no longer always be sustained under new economic pressures. The pre-existing variability in processes of household formation and division among the Burkinabe Fulani certainly helps to facilitate new diversified livelihood strategies such as exode.
But, in addition to simply taking advantage of household division being delayed for other reasons, there were at least four cases of men deliberately delaying this process in order to continue reaping the benefits of exode without undermining domestic agro-pastoral production. This flexibility, which pre-dates exode, means that it is possible to manipulate processes of household division in order to allow new strategies, such as exode, without challenging the accepted norms governing the domestic cycle.
An example of this is what subsequently happened in case C above, in which a Fulbe Gaobe household looked set to split: following his marriage, the babaade's eldest son decided that he was no longer prepared to put any of his millet in the granary of his mother's co-wife. He and his younger brothers built their own granary, into which they put millet for their mother alone, effectively forcing the sons of the co-wife to do the same. When I wrote this account, at an early stage in fieldwork, it looked as if a split was imminent. However, as fieldwork progressed, it became clear from further discussions with the young man that, although he would like to take the division further: to become a babaade in his own right and to set up a separate wuro, this was unlikely to happen immediately, despite the posturing to that effect. He spent every dry season in Abidjan, as did several of his younger brothers, and he was concerned about leaving his mother and wife alone in a small wuro, with no men around during his absence. For this reason, he was delaying leaving his father's wuro, and would probably stay until he, or one of his brothers, was more settled and could stay permanently.
An account by a young Pullo Liptaako tells a similar story. Tensions between brothers' wives were pushing in the direction of separation of the baade. But, at least at the time of fieldwork, the brothers had resisted separating their herds and fields, so that some of the brothers could leave en exode while one or two stayed and looked after the whole baade.
Where, for whatever reason, a baade or wuro is too small to leave a viable production unit while men are away en exode, other strategies may be employed to overcome shortfalls in labour. One such possibility is the fusion of two previously independent units, on either a permanent or a temporary basis. In practice, the permanent coupling of baadeji/ngure is almost unknown. The only normal circumstance under which a man stops being a babaade is by becoming too old and incapacitated; even then, he often retains the nominal title of babaade.
However, the temporary joining of smaller baadeji/ngure to accommodate the seasonal absence of migrant men is not uncommon. This has been observed elsewhere (Maliki et al. 1984; David and Niang 1995). Of the 117 baadeji/ngure sampled in the multiple round study, 41 sent men en exode during the dry season 1995-6, and eight experienced the temporary absence of all adult men through exode. Of these eight, four joined with other households during the men's absence. In one case, a young married woman whose husband was away simply rejoined her parents' wuro: in other words, the migrant's wuro was temporarily absorbed by another during his absence. Another migrant's wife joined her married sister's wuro, and a third joined the baade of her mother-in-law's brother. In the final case, three wuro fragments combined, the absent babaade'en's mothers being sisters.
A noteworthy point is that the temporary attachments are typically not between groups of people that might, under other circumstances, have been part of the same baade/wuro. Instead, they draw on matrilateral kinship relations that fall outside the patrilineal descent-based kin groups. Women spend considerable amounts of time paying extended visits to their own family members. It can be seen from Table 1 that 11.7 per cent of women in the MRS sample visited friends and family outside of their wuro/baade for at least one week in the study year--and most of these visits were for considerably longer than a week. A further 6.7 per cent of women went back to their parents' home to give birth, staying typically for several weeks, and even several months for first births. Regular, extended visits to family and friends are useful for consolidating ties, which can be drawn upon in times of trouble, and indeed to maintain domestic viability in the face of temporary absences of men. Bonds between Gaobe women in particular are also created and reinforced during trips to the Mossi Plateau to buy grass for weaving nattes, used for making tents (cuudi) (Table 1). These trips, which may last several weeks, are undertaken in dabbunde (the cool, dry season) and represent an important occasion for groups of married women to be together, away from their menfolk.
Again, this sort of flexibility is not a new phenomenon, nor is it linked exclusively with exode. Oral histories indicate that the fluidity of household boundaries and the temporary joining of ngure/baadeji have long been used to overcome labour shortages, particularly those associated with herding and transhumance. In the 1996 study, there were five further cases of baadeji/ngure combining for this purpose, with three more joining briefly during the absence of women, for the purposes of food preparation. In other words, like flexibility in household division, the temporary joining of households is not a function of recent exode. Rather, it can be used to facilitate exode without needing to resort to new forms of domestic organization.
An alternative to the temporary incorporation of one household into another is the use of less formal extra-household support networks to plug labour deficits left by migrants. This is a common strategy particularly when exode removes some, but not all the men, from a baade/wuro. As exode takes place predominantly during the dry season, this affects primarily herding work, rather than agriculture. Of the 34 baade/ngure involved in pastoral production that sent migrants over the dry season 1995-6, 13 relied on kin networks beyond the household to help with herding work, to ensure that domestic production did not suffer as a result of the exode. These included two of the eight temporarily man-less households that were not incorporated into other units. Again, matrilateral ties were important: seven of the 13 drew predominantly on these.
IMPACTS OF OUT-MIGRATION OF MEN ON INTRA-HOUSEHOLD GENDER DIVISIONS OF LABOUR AND POWER
The preceding discussion indicates that a good deal of flexibility exists in domestic organization among the Burkinabe Fulani, which serves to maintain viable units of domestic production and consumption in the face of the temporary absence of substantial numbers of young men. This flexibility means that the normal processes of production and consumption can continue relatively undisturbed in the face of exode. As a result, there is little need for changes to gender divisions of labour and power.
There is a very clear gender division of labour among the Fulani of Burkina Faso, which is also found among other Fulani populations (Hopen 1958; Stenning 1958, 1959; Dupire 1963, 1970; De Bruijn and Van Dijk 1995; de Bruijn 1997). Men are responsible for most agro-pastoral production tasks and for providing the baade/wuro with basic foodstuffs: grain, milk and meat. These tasks include: (1) for pastoralists: herding cattle, and taking them to pasture and water; (2) for agriculturalists: preparing the fields for cultivation, sowing the crop, hoeing (the most labour-intensive task) and harvesting. Women's tasks revolve mostly around preparation of food, childcare, and other domestic duties, such as keeping the suudu and its surroundings clean and tidy. While men are responsible for providing the basic foodstuffs, it is up to women to provide condiments (spices, leaves, soluble seasoning cubes) that make the food more palatable. The only tasks associated with agro-pastoral production that are performed regularly by women are thinning the crop of millet and, for Fulbe Djelgobe women only, milking cattle.
This gender division of labour is strictly adhered to in practice. During 14 months of intensive fieldwork, the instances I observed of people breaking these unwritten rules were rare and always attracted comment, and often derision. This usually involved an older woman having to do some work that her sons should have been doing, but were not, because there were said either to be too lazy or too incompetent (see below). Moreover, there was very little variation between lenyi/status groups in the gender division of labour. As noted above, one exception to this was that Fulbe Djelgobe women regularly milked cattle. Among all of the groups, women owned and could decide what to do with the milk, which provides an important source of independent income for women. However, apart from the Djelgobe case, the women were given milk by their husbands or other menfolk, rather than being allowed to milk cattle themselves (the men claimed that, if it was left to the women, they would take too much milk, leaving insufficient for the calves). As observed also by Bolwig and Paarup-Laursen (1999), the Riimaaibe Liptaako in this area observe the same practices of gender division of labour as the Fulbe, with women refraining from undertaking tasks (other than thinning the millet crop) associated with agro-pastoral production.
It is because of the strict gender division of labour that it is so crucial to keep at least some adult men during the period when others go en exode. Because exode happens over the dry season (dabbunde and ceedu), agricultural tasks are fairly light: all that is required is preparing the fields for cultivation. For those with livestock too, however, the work is more demanding, as discussed above: pasturing and watering become more labour-intensive at this time of year.
Maintaining large ngure/baadeji, partly through delaying the processes of domestic division, means that, in the majority of cases, exode does not rob a household of all of its adult men. Only eight of the 41 baadeji/ngure sending migrants in 1995-6 were left entirely bereft of adult men during the dry season. In the other 33 cases, there was at least one man left to perform 'male' tasks and responsibilities. Where necessary, extra-household networks were drawn upon to assist. In none of these 33 baadeji/ngure did the women take on any of the productive tasks usually associated with men. Nor did they assume any new decision-making powers over agro-pastoral production; all such decisions were taken by the remaining men. If anything, the decision-making power of migrants' wives within the household was reduced. Without their husbands to speak up for them, several women whose husbands were away complained of having to shoulder an unfair share of tasks such as pounding millet and fetching water, and of