Chapter Eight—
Labor Circulation, Marriage, and Fertility in Southern Africa

Ian Timaeus
Wendy Graham

Introduction

Throughout Southern Africa, circulation of members of the labor force between rural areas and the major centers of wage employment is an important and long-established feature of people's lives. In some areas, such as the "homelands" of South Africa, Lesotho, and parts of Botswana, labor migration involves a large proportion of the adult male population and affects nearly every household. Although such movement is usually temporary, it frequently entails the prolonged separation of migrants, especially young men, from their immediate families and other relatives. The effects on relationships between close kin, on marriage systems, and on household structure have been wide ranging and to some extent are only now becoming apparent. Equally the pattern of labor migration has been influenced by, and integrated with, the domestic institutions and economies of the indigenous populations of the region as well as by wider economic and political relationships and inequalities. Thus, in order to fully understand the reproductive regimes of this part of Africa, it is essential to consider them in the context of the migrant labor system and the political economy of the region as well as in their local and familiar institutional framework.

One aspect of labor circulation with which we are concerned is its direct effects on the proximate determinants of fertility. By direct effects we mean those aspects of labor circulation that influence only the fertility of migrants and their sexual partners. Such effects may have an immediate impact on fertility, for example, through the separation of spouses at times when the woman would otherwise be exposed to the risk of conception, or a longer-term impact, for example, through women whose husbands work as miners having an increased chance of being widowed at a young age. We also

consider the indirect impact of labor circulation on the fertility of entire populations. Contextual effects on the domestic economy and attitudes to reproduction are important. In addition, in the longer term, the impact of labor circulation on the development of social institutions has to be considered. In particular, we argue that the economic opportunities and roles for both men and women associated with widespread migratory wage employment have contributed to the constitution of distinctive marriage systems. The main features of these systems are documented in chapter 6 of this book. Throughout the region, ages at marriage for both sexes are late and the prevalence of polygyny is low compared with other sub-Saharan African countries. Moreover in Botswana a growing proportion of women never marry. Marital breakdown has become increasingly common and divorce, separation, and widowhood are seldom followed by remarriage for women. These trends help explain changes in and the increasing diversity of the structure of domestic groups and, in particular, the emergence of households headed permanently by women.

The discussion that follows commences with a brief description of the development of the migrant labor system within and between the countries of Southern Africa. Against this background we discuss the marriage systems of these countries and the effect of labor circulation on each of the determinants of the proportion of women of childbearing age who are living in conjugal unions. It is the household that is the locus of the labor circulation and marriage systems. The developmental cycle of domestic groups forms a framework that shapes, and is shaped by, its members' migrant careers, marital histories, and fertility. Following our analysis of these interrelationships, we examine the direct impact of labor circulation on levels of fertility. Finally we discuss the extent to which the migrant labor system is related to the adoption of modern methods of contraception in the region, and we assess the prospects for fertility decline. Throughout, the discussion concentrates on Botswana and Lesotho and on the large majority of their inhabitants whose permanent homes are in the rural areas. In part this reflects the authors' familiarity with these countries and in part the availability of data. Where appropriate we draw on surveys conducted in South Africa and Zimbabwe but, while aspects of the discussion may be relevant to these countries, their massive internal migration flows undoubtedly have different demographic implications from the international movements on which we concentrate. We would also be very cautious about extending our arguments to the matrilineal kinship systems and different domestic economies characteristic of much of Zambia and Malawi.

The Migrant Labor System

The populations of Southern Africa are frequently described as "migratory societies," and an extensive literature exists on all aspects of this mobility—

from its historical roots to its pervasive impact on the politics, economics, and social structure of these countries (for example, Bell, 1986; Kowet, 1978; Magubane, 1975; Mitchell, 1985; van Binsbergen and Meilink, 1978). The presentation of a profile of migration for this region of Africa is complicated by the commutation of determinants and consequences, by the interplay of historical and contemporary forces, and by the variety of types of movement grouped under the term migration. We therefore limit ourselves to introducing important dimensions of the patterns and processes of migration in Southern Africa that are relevant to fertility.

It is important to begin by clarifying a couple of terms. Much of the literature on mobility makes a basic distinction between migration and circulation (Prothero and Chapman, 1985). The former implies some form of permanency—or intention to stay permanently—at a destination, while the latter describes short-term, repetitive movements. While this distinction is difficult to adhere to strictly, since migration is also the term used colloquially to describe mobility in Southern Africa, it is important to emphasize that the majority of movement with which we are concerned is circulation. It is the historical development of this circulation and the political, economic, and social setting in which it takes place that helps to explain its character.

There are several major works on the origins of the migrant labor system (see, for example, Bundy, 1979; Cliffe, 1977; Parsons, 1977; Schapera, 1947). The early evidence suggests that although migrations took place in the traditional societies of Southern Africa, often related to season or to tribal conflicts or environmental pressures, an increase in the volume and diversity of movements can be linked to contact with Europeans. Several authors have indicated that the initial migrants were age regiments of young men sent by the local chief to work on nearby settler farms in exchange for guns and other European commodities. By the midnineteenth century, however, much of Southern Africa was being absorbed into a monetary economy in which cash was required to meet new needs and to pay the taxes imposed by European administrations. Many of the migrants were young single men who oscillated between their parents' homesteads and nearby farms. Indeed a period of earning cash became a form of rite de passage into manhood. Although young women were reported to have worked as agricultural laborers or domestic servants on European farms, this was comparatively rare. Indeed there is evidence that chiefs and elders often prohibited the outmigration of women (Molenaar, 1980). It is hard to generalize about these early movements as they took a great variety of forms. Nevertheless, in the areas covered today by Botswana, Lesotho, South Africa, and Swaziland, it was primarily individuals who migrated while, in the regions of white settler agriculture in Zimbabwe and Zambia, whole families tended to move to take up employment.

Although farm work played an important role in the initial development of the migrant labor system and remains numerically significant in areas

such as the Transvaal and southern Botswana (Wylie, 1981), it was the establishment of mining and manufacturing industries that provided the major stimulus to the system's growth. Exploitation of the unskilled labor of temporary migrants from neighboring territories and distant tribal areas has been the basis of South Africa's industrial and commercial growth for a very long time. The practice can be traced back to the opening of diamond mines in Kimberley in the 1870s and of gold mines on the Witwatersrand from the mid-1880s, but it was the early part of this century that saw a massive upsurge in numbers migrating as can be seen from table 8.1. The interplay of environmental and economic factors contributing to this trend is well documented in the literature (for instance, Colclough and McCarthy, 1980; Leggasick and Le Clerq, 1978; Palmer and Parsons, 1977; Skinner, 1985). Although the precise combination of factors varied over time and between populations, among the most influential were: the lack of local wage-labor opportunities; recurring droughts and cattle disease; serious land shortages following appropriations for white settler farming; the introduction of the hut tax, tribal levies and fines; and the destructive effect on the domestic economy of the loss of manpower and skills that arose from outmigration itself. Although an extensive catchment area developed, with workers being drawn to the Republic of South Africa (RSA) from as far north as Malawi and Northern Mozambique, it was the territories abutting the RSA that came to be regarded as labor reserves. Even within this area, however, there have always been regional and ethnic variations in the extent of labor migration. Some recent figures are presented in table 8.2. In Lesotho as a whole in 1975, 60 percent of the male labor force was employed in the RSA and 81 percent of them were working in the mining sector; 90 percent of men had worked in the RSA at some point in their lives (van der Wiel, 1977). In Botswana it has been estimated that, until the mid-1970s, 50 percent of those employed in the RSA worked in the mines, 10 percent in agriculture, and the rest in manufacturing and the service sectors (Cooper, 1979; Kerven, 1979a ).

The complex process by which migrant labor in Southern Africa was institutionalized by the labor-recruiting agencies and developed into a self-perpetuating system has been extensively studied (for instance, Clarke, 1977; Mitchell, 1985; Parsons, 1977). The social and economic interdependence of migrant workers and their rurally based kin is crucial to the maintenance of the system. Low mine wages and South African policies of apartheid and separate development have restricted movement to adult laborers and excluded their families (Colclough, 1980; Taylor, 1982; Wilson, 1972). Conversely loss of manpower in the rural areas reduced the productivity and viability of farming and helped to create rural dependency on migrants' remittances. Kin in the areas of outmigration, in turn, provide for the longterm security of the migrants, maintaining rights to land and raising the next generation of the labor force.

One aspect of the migrant labor system that is relevant to the discussion that follows is the arrangements operated through the Chamber of Mines for the recruitment of foreign labor. Mine workers are employed on fixed-term contracts, usually of 9 months duration, and must return to their home country on completion of their contract even though they usually take up another one a few months later. Circulation between the laborers' homes and the mine dominates the working lives of these men and their families. In these circumstances, household structure is in a constant state of flux, with marked differences existing between de jure and de facto household composition and headship. For example, in Botswana over 10 percent of households are headed by women whose husbands are temporarily absent (Izzard, 1981).

Schapera (1947) provides some of the earliest insights into the social implications of labor circulation and many of his observations remain relevant today. Writing primarily about the Tswana, he noted that 90 percent of male absentees were between 15 and 44 years old and that they spent about 12 years of their working lives in the mines. A mine contract often marked entry into manhood, and migrants continued to take up contracts until their sons were old enough to migrate themselves. At this point they could expect some support from their sons and tended to retire from wage employment to their rural homes.

The characteristics of the dominant migrant streams are reflected in the demographic structure of the areas of outmigration, which have high dependency ratios and a marked predominance of women—often referred to as the "women left behind." There is an extensive literature on the role these women play in the maintenance of the migrant labor system and on how the overall status of women has developed in response (Bozzoli, 1983; Brown, 1979; Gay, 1980; Gordon, n.d.: Izzard, 1982; Kooijman, 1978; Meillassoux, 1975; Murray, 1981; Wolpe, 1972). As mentioned previously, rurally based kin offer an incentive and opportunity for migrants to return, and their value is implicitly recognized in the system of Voluntary Deferred Pay that operates to ensure that miners remit part of their wages to these kin. In Botswana, such remittances may contribute nearly half of the total income of a family as well as injecting cash into the rural economy (Kerven, 1982; Lucas, 1982). On the other hand, constraints on entry, residence, and movements in the RSA, and insecurity of employment arising from withdrawal of contracts, redundancy, or ill-health, require migrants to maintain strong contacts with their place of origin. However, the links between a migrant and his kin are not solely economic. Social obligations are also important, particularly with regard to the interdependence of generations and expected patterns of support in old age.

Although the pattern of labor migration in Southern Africa at the end of the 1970s was in many ways similar to that observed for the previous 50 years, more recently there have been a number of significant changes. Two of them require brief discussion; first, the change in the pattern of mine labor recruitment, and second, the growth in opportunities for wage employment within some of the labor-exporting countries. Table 8.3 reveals the expansion in the recruitment of black South Africans for mine work between 1973 and 1985 and the concomitant decline in the employment of foreign workers. A number of authors have identified similar falls in the demand for foreign labor in other sectors of the RSA's economy (Clarke, 1977; Davies, 1978; Wylie, 1981). To some extent the fall in the employment of foreigners in the RSA has been offset by the growth of employment opportunities in the black African states, in particular Botswana and Zimbabwe. This upsurge in formal sector employment is a postindependence phenomenon and is concen-

trated in urban areas. In Botswana the development of the mining sector, in particular, led to a 53 percent increase in the number of people in formal employment between 1972 and 1976. Although it is estimated that between a quarter and a half of all rural Batswana households still have at least one member working in the RSA, the emergence of internal townward migratory flows is an important development. Two of its features that are relevant here are the continuing importance of circulatory movement and the comparatively high proportion of female migrants involved.

So far the discussion has concentrated on male migration. While female labor circulation is numerically less significant, as indicated in table 8.4, it is an important indicator of the changing social and economic roles of women in Southern African societies (Bryant, 1977; Chaney, 1980; Izzard, 1981 and 1985; Marks and Unterhalter, 1978; Murray, 1981). In the earlier part of this century outmigration was a means of 'escape' for women and was often a response to extramarital pregnancy, infertility, or marital breakdown. In-

creasingly, it has become part of the diversification of women's roles to encompass the positions of breadwinner and household head. The predominant occupation of women in the major centers of employment is as unskilled workers in the service sector and, in particular, as domestic servants (Cock, 1980). Three-quarters of the Basotho women interviewed by Gay (1980) who had worked in South Africa had been employed as servants. The equivalent figure from a study in Botswana was 57 percent (Izzard, 1982). While the unregulated nature of employment in domestic service has enabled many women to avoid some of the tight controls on entry and movement experienced by mine workers, they have suffered from the absence of standards concerning hours of work, wages, job security, and accommodation. Many female migrants have to supplement their income with informal activities such as beer brewing and must delay or delegate to others their role as mother. Insecurity of employment and multiple roles and activities lead women, like men, to maintain strong links with their rural kin and to circulate, throughout their working lives, between the centers of employment and their rural home. While their pattern of movement may resemble that of men, the distinctive feature of rural–urban interdependence for women is its association with the development of family forms in which the role of the father is marginal in both social and economic terms and of households headed permanently by women (Gay, 1980; Izzard, 1982). In Botswana the growth of local employment opportunities has affected women as well as men. Townwards migration has been dominated by women since the mid-1970s, although rising domestic formal-sector employment of Batswana women has been roughly paralleled by a fall in numbers working abroad (Izzard, 1982).

The Marriage Systems of Botswana and Lesotho

In this section we discuss the marriage systems of the Basotho and Tswana in their wider social context. We speak of marriage systems to emphasize that each aspect of marriage patterns, for instance, ages at marriage or proportions widowed, is related to all the others. This is exemplified by the discussion of polygyny in earlier chapters of this book. Similar considerations apply to the interrelationships between nuptiality and fertility. For example, while polygyny probably lowers the fertility of young wives, it may also be linked to the remarriage of a higher proportion of older widowed and divorced women (van de Walle, 1985). Moreover, when discussing the historical development of marital institutions, it is probably misleading to consider modification of any single aspect of the marriage system as the cause of changes in its others. Rather explanations should be sought in wider circumstances that lead people to alter their patterns of behavior in ways that gradually affect all aspects of marriage. Having said this, we discuss each facet of marriage patterns separately in order to impose some structure on the argument.

An apparent contradiction in the development over the last 100 years of the marriage systems of Botswana and Lesotho has been the continuity of the basic structure of marital institutions despite massive changes in marriage patterns and disruption of family life (see, for example, Kuper, 1985; Molenaar, 1980; Murray, 1981). A brief outline of some of these structural aspects of marriage is essential to the discussion that follows. While lineages are not corporate property-owning groups, residential groups have always been and remain constituted according to patrilineal premises. In line with these, residence on marriage is virilocal, that is to say that the woman moves into her husband's household or that of his parents (Preston-Whyte, 1974). The jural component of marriage is still the transfer of bridewealth reckoned in cattle in exchange for rights in the woman's children. Marriage without cattle remains very rare, at least in rural areas, but the payment of bridewealth may only be completed years after the initiation of the sexual and domestic components of the marital relationship. Therefore getting married is a protracted process rather than a single event (Comaroff and Roberts, 1977). As a result, in the event of marital dispute or disruption, people's marital status may become hard to define. A man is expected to provide a dwelling, agricultural land, and cattle for his wife to support her and her children (Poulter, 1976). In a polygynous union each wife has to be provided for separately. Inheritance and the circulation of bridewealth transfers are concentrated within a woman's "house" (Preston-Whyte, 1974). She forms the nucleus of a set of patrimonial property and the pattern of marital relationships defines rights over productive wealth (Gluckman, 1971).

Polygynous marriages are now rare among the Basotho and Batswana. According to the 1977 Lesotho Fertility Survey (LFS) only 8.5 percent of married women were living in polygynous unions and only 1 percent of them had more than one cowife (Bureau of Statistics, 1981). The resilience of polygyny in the face of social and economic change in most of sub-Saharan Africa has been documented in this book (see chapter 6). In Southern Africa the practice was a far less durable feature of the marriage system. In Botswana, nobles initiated between 1830 and 1860 married an average of 3.3 women and commoners 1.9 women. Subsequently such marriages became less and less common and men initiated after 1920 married an average of only 1.1 women. (Kuper, 1985). There are no nineteenth-century data for Lesotho, but census statistics for the early part of this century reveal that the proportion of married men with more than one wife fell from 18.7 percent in 1911 to 8.9 percent in 1946 (Murray, 1981). These developments occurred so early that it is probably futile to attempt to assess the relative importance of the factors underlying them. The list of those that were probably relevant, however, seems fairly clear. The influence of the Christian missions was one such factor. It affected not only individual converts but also underlay attempts at the legal abolition of polygyny in Botswana in the early part of this century. Second, polygyny was integrated with traditional relationships

of clientage whereby chiefs and other political figures, who controlled rights to land and citizenship, amassed wealth in the form of cattle and the productive and reproductive capacities of women. Polygynous marriages enabled better-off men to enlarge the size of the kin group over which they had influence. The diffusion of political power, initially to colonial officers and later to elected politicians and official bodies, led to the breakdown of these relationships of clientage. As a result traditional political leaders lost the opportunity to express and reinforce their position by accumulating wives and offspring (Kuper, 1985; Murray, 1981). Third, the importance of women's contribution to agricultural labor is one factor closely associated with the prevalence of polygynous marriage in sub-Saharan Africa (see chapter 6 and Boserup, 1970). The period of colonial expansion led to the confinement of the Basotho in a mountainous and increasingly overpopulated area and of the Tswana in one much more suited to pastoralism, a male activity, than to farming. Agriculture became less viable at the same time as new opportunities for wage employment developed. In addition, the adoption of the plough in the late nineteenth century increased men's contribution to agricultural work. As a result, polygynous families became less economically advantageous for men in comparison with the cost of bridewealth and the difficulty of acquiring an independent provision of land for each wife (Kuper, 1985; Murray, 1981 ).

Ages at marriage are high for both sexes throughout almost all of Southern Africa. For Lesotho, the LFS household data shown in table 8.5 yield a singulate mean age at marriage of 20.1 years for women and 24.8 years for men (Bureau of Statistics, 1981). Nearly 70 percent of women marry by age

25 and 95 percent by age 30 (Timaeus and Balasubramanian, 1984). Late marriage is not a recent development. A detailed assessment of the LFS marriage histories and census data suggests that for women it dates back to at least the 1940s. In fact ages at marriage declined slightly between the mid 1960s and mid 1970s (Timaeus and Balasubramanian, 1984). Moreover anthropologists conducting fieldwork in the 1930s reported that men married at between 23 and 26 years and women at ages 18 to 24 (Ashton, 1967). In Botswana the pattern of first marriage is different. Village studies conducted by Schapera in the 1930s and 1940s suggest that ages at marriage were already very high, in their early twenties for women and their late twenties for men (Kuper, 1985), and appreciable numbers of women never married (Molenaar, 1980). The picture revealed by the 1971 Census data shown in table 8.5 is even more striking. The singulate mean age at marriage is 24.3 for women and 29.3 for men (Central Statistics Office, 1972). Moreover some 17 percent of men and women aged between 35 and 50 have never married.

In view of the link between female education and marriage patterns throughout sub-Saharan Africa (see chapter 6), one factor that may be related to late marriage is the high level of schooling of the populations of Botswana and Lesotho. The Christian missions were responsible for the early development of the school system and the LFS data on older women suggest that by the 1920s and 1930s a majority of girls in Lesotho were receiving some formal education. Today about 95 percent of girls attend school. In Botswana around 30 percent of girls attended school in the prewar period, rising to about two-thirds more recently (Central Statistics Office, 1972). Notably, fewer men than women receive elementary education. In Lesotho, about 30 percent and in Botswana a majority of boys never go to school. The importance of labor migration is one explanation of this pattern. On the one hand mothers and, in particular, female household heads favor the education of their daughters as they are more likely to remain attached to their natal households than sons and as knowledge of English is often required of domestic servants (Allison, 1979; Gay, 1980). On the other hand, education is not essential for mine work and the shortage of male labor means that boys are often removed from school to herd cattle.

Considering that the pattern of late marriage had already been established in Botswana and Lesotho by the 1930s and 1940s, it is also plausible to link its development to the decline in polygynous marriage. Other things being equal, a decline in polygyny will bring about delayed entry into marriage for women. Moreover the economic disincentives to polygynous marriage for men would also have discouraged them from marrying at young ages. Today men who lack other sources of income can only afford to marry after they have spent one or more periods as migrants earning a cash income. In Lesotho bridewealth transfers are high and are no longer drawn from a wide variety of a man's kin and distributed among his wife's relatives (Mur-

ray, 1981). They now represent a transfer between two households and are largely funded by the cash earnings of the man concerned. Furthermore, Basotho men aim to establish an independent household as soon as possible after they marry (Gay, 1980). One major expense that this involves is the acquisition of a dwelling, but it also implies access to agricultural land and cattle as the household's subsistence basis. Arable land in both countries is vested in the nation; it cannot be bought or sold. Rights to cultivate land are a prerogative of citizenship and are allocated to household heads by chiefs advised by land committees in Lesotho (Murray, 1981) and by landboards in Botswana (Molenaar, 1980). In theory they are allocated on the basis of need and ability to use the land and there is some expectation that men will gain access to their fathers' fields. In practice, particularly in Lesotho, the shortage of land makes obtaining rights to it difficult. Influencing the land committee in one's favor involves time, effort, and money (Murray, 1981). Moreover successful agricultural activities now require investment that most households can only fund by migrant earnings (Molenaar, 1980; Murray, 1981). Thus late male marriage, leading to late female marriage in the absence of polygyny, is at least in part a function of the interrelationship of labor migration and the domestic economy. To quote one of Murray's informants (1981, Sesotho translations omitted):

On your first trip to the place of the whites you support those who brought you up. On your second trip you take out money that counts as cattle for marrying a wife. On your third trip, you look after everything in your own homestead.

In line with this analysis, the slight decline in ages at marriage between the mid-1960s and mid-1970s coincided with a period of rapid growth in mine earnings that would have made it easier for men in their twenties to marry.

Despite the economic obstacles to marrying early, most Basotho do establish a conjugal relationship at a fairly young age, whereas in Botswana marriage often occurs at much later ages. Kuper (1985) has suggested that this is linked to the different rural economies of the two countries rather than to variation in their cultures. As we have already outlined, the insecurity of wage employment and restrictions on permanent migration encourages migrants to maintain close links with their place of origin. Their long-term security depends on investment in the rural economy and the next generation. In Lesotho marriage is essential to this (Gay, 1980). Only married men can gain access to land and land has to be cultivated to be retained. In general, a man has to have a wife to acquire and maintain a farm for his old age. In contrast, in Botswana pastoralism is the central component of the rural economy. Agriculture is far less important than in Lesotho, and labor migration has never become as common. In order to provide for their old age men are likely to invest in cattle. The rural economy is stronger than in Lesotho and some men continue to exploit the opportunities offered by marriage for investment

in internal political and economic enterprises. They tend to marry at young ages and often contract polygynous unions (Comaroff and Roberts, 1977). Migrant workers, however, often remit funds to their fathers or brothers who will buy cattle for them and tend them with their own herds. In this situation there is little incentive to marry before retiring from wage employment (Gulbrandsen, 1986; Kooijman, 1978; Kuper, 1985).

Another aspect of marriage patterns of significance for fertility and women's life courses is the high level of female widowhood in both countries. As can be seen from tables 8.5 and 8.6, appreciable numbers of young women are widows and the majority of women who marry are widowed before they reach old age. This pattern results from the abnormally high level of male compared with female mortality in both countries, from differences in ages at marriage, and from the infrequency of remarriage for widowed women. At the mortality rates prevailing in Lesotho in the mid-1970s, 35 percent of 25-year old men would die before age 55 while, in contrast, only 14 percent of women aged 20 would die before their fiftieth birthday (Timaeus, 1984). In Botswana adult male mortality is lower but still unusually high compared with the level of adult female mortality (Central Statistics Office, 1972). Unfortunately it is impossible to unravel with the data available either the exact contribution of labor migration to excess adult male mortality or the history of mortality differentials. However, a wealth of evidence attests to the appalling impact of minework on the health of the labor force earlier this century, and to the continuing dangers of accidents at work and occupationally related disease (Doyal, 1979; van Onselen, 1976).

Traditionally divorce was very rare in Botswana and Lesotho. A prolonged debate within social anthropology about the determinants of divorce rates in African societies concluded that this was linked to the inheritance system. The property allocated to a married woman's "house" passes to her sons and divorce has a very disruptive effect on rights over property (Gluckman, 1971). Both the families of an estranged couple and the traditional courts exert considerable pressure on them to maintain their relationship (Ashton, 1967; Schapera, 1940). However, as can be seen from table 8.5, marital breakdown is now quite common, although much less prevalent than in the West African countries surveyed by the World Fertility Survey (McDonald, 1985). Moreover it is likely that statistics gathered in large-scale demographic enquiries underestimate the extent of marital disruption. The hostile attitude of the courts means that marriages often end in separation rather than formal divorce, and some women who are still living in their marital home have been effectively deserted by their husbands (Gay, 1980). For example, the LFS household data show that 6 percent of women who said that they were married did not have a partner associated with their household. In addition, as we have outlined, the process of marrying someone is only completed after the entire bridewealth transfer has been made. If

marital disruption occurs before this point, respondents may redefine themselves as not having been married, even if the domestic and sexual components of the relationship had been well established, and fail to report the union. On the one hand, the economic opportunities and roles associated with wage employment and labor migration have weakened the social sanctions that used to inhibit marital breakdown. On the other hand, prima facie one would expect the prolonged separation of spouses stemming from labor migration to increase the likelihood of marital breakdown. Ethnographic studies support this conclusion. As Murray (1981) argues:

A man's absence as a migrant labourer is a condition of his family's survival. But his absence also undermines the conjugal stability from which his family derives its identity.

Women have a heavy responsibility for household management in the absence of their husbands but are dependent on reliable cash remittances. Not suprisingly disputes about how money is spent are common. Moreover women regard it as inevitable that their husbands will have affairs, especially when they are working away from home, and women often have affairs themselves (Gay, 1980).

The other factor contributing to the large number of widowed, divorced, and separated women is the exceptionally low level of female remarriage. The LFS marriage histories indicate that 15 percent of ever-married women of childbearing age had experienced the dissolution of their first marriage but only 29 percent of divorced and separated women and 5 percent of widows remarried within 5 years (McDonald, 1985). In part this reflects demographic factors. As fewer men now contract polygynous marriages, excess male mortality and the more or less permanent emigration of a minority of men make it difficult for an older woman to find a new husband. Social structural factors are also important and help to explain why almost no widows remarry. In jural terms widows remain under the authority of their husband's (and sons') agnatic kin. Traditionally they would have become the wife of one of his close relatives, ideally his younger brother (Ashton, 1967; Schapera, 1940). Like divorce, remarriage of widows raises problems about the group membership and inheritance rights of the next generation. Poulter (1976) reports that most widows in Lesotho believe that remarriage is illegal, although the actual legal position is less clear-cut. In addition, a widowed woman has rights to, and effective control over, the building, fields, and property that, together with her offspring, constitute her "house" (Gay, 1980; Kerven, 1979b ; Poulter, 1976). She can also expect support from her affines in farming her land. In Botswana, they will tend her cattle for her, although this often leads to disputes about ownership with widows claiming that their husband's brothers have "eaten" their cattle (Molenaar, 1980). Many widows experience great hardship. Nevertheless, if they remarry, they

lose both their independence and control of the productive wealth that they manage on behalf of their dead husband's heirs.

The marriage systems that we have described are intimately interlinked with the developmental cycle of domestic groups and shape the household circumstances of women of childbearing age. Moreover, as we have outlined and as a number of workers have elucidated in detail, migrants' careers and the developmental cycle are also closely interrelated (Izzard, 1982; Mitchell, 1969). In Lesotho, Gay (1980) has outlined six common domestic situations that women are likely to experience in the course of their lives. The first is that of the single woman who remains subordinate to her parents even if she resides elsewhere. The second is a usually brief period spent as a married woman in her parents-in-law's household. The third is married life in her husband's household acting as the de facto household head while he works in RSA. The fourth occurs when her husband retires and returns to live with her permanently, and the fifth when he dies and she becomes a household head in her own right. Finally there are divorced and separated women who usually become reattached to their natal home. Clearly this model ignores the complexity of some women's lives. However, although one must be careful about inferring longitudinal experience from cross-sectional data, table 8.6 reveals the importance of the categories distinguished by Gay and the way in which women's domestic circumstances are likely to change during the course of their lives. Thus the structure of the households that women live in are liable to change as the wider domestic group responds to changing opportunities and constraints. Migration is one such response. The domestic situations described by Gay are also characteristic of women's experience in Botswana (Izzard, 1982). What is distinctive about the current situation there is that household fission now occurs at a far later stage in the developmental cycle than in Lesotho. In the ward studied by Molenaar (1980), average household size increased from 5.6 people in the 1930s to 10.4 in the 1970s. National data reveal that the mean de jure size of households is 9.5 (Izzard, 1979), compared with an equivalent figure of 4.9 in Lesotho. This reflects two facts. First, delayed marriage in Botswana means that large numbers of adult men and women now remain members of their natal household till quite advanced ages. While many of them will be absentees, children of a migrant woman will usually be fostered with her kin. Second, the relative importance in Botswana of cooperation among men in cattle herding, as opposed to setting up an independent household, means that brothers often continue to live together after they marry and even after the death of their parents. Male migrants in particular frequently delay setting up their own household until they retire from wage employment (Molenaar, 1980). As a result women are less likely to spend a lot of their adult lives as de facto household heads in Botswana than in Lesotho. On the other hand, while the majority of de jure female household heads in Lesotho are widows, in Bots-

wana older single women with regular employment or economically active children also tend to establish their own households (Izzard, 1985). In both countries most women spend at least some part of their lives as household heads.

Marital Status and Fertility

The relationships between marital status and lifetime and current fertility are examined in tables 8.7 and 8.8. The impact of late marriage, frequent marital dissolution, and low rates of female remarriage on fertility is less than one might expect. Although many women of reproductive age are not formally married, childbearing outside marriage is common. The Botswana Family Health Survey conducted in 1984 enquired about informal unions (Manyeneng et al., 1985).^[1] As can be seen from table 8.9, the responses reveal that many women have informal relationships. It would be misleading to consider these relationships as socially equivalent to marriages. The ethnographic literature emphasizes the transitory and exploitative nature of many of them (Molenaar, 1980) and often they do not imply membership of the same household. According to the LFS household survey, 16 percent of widowed women of childbearing age do have a partner who is a member of their household. However, recalculation of the current fertility rates reveals that whether or not women are living with a partner has even less of an effect on their fertility than their marital status. Ever-married women who were not living with a man had 94 percent of the number of births in the preceding year of an equivalent group of women with partners in the household.

The level of premarital fertility is markedly different in the two countries.

Single women in Lesotho have fewer children than those of comparable ages in Botswana and had only 13 percent of the births in the year before the survey that one would have expected if they had been married. Moreover, according to the birth histories collected in the LFS, only 5 percent of women had a birth before they married and only another 5 percent were pregnant when they married (Bureau of Statistics, 1981). In Botswana, according to the preliminary results of the 1981 census, 7 percent of single women aged 15–19 and 14 percent of those aged 20–24 have a birth each year (Lesetedi, 1984 quoted in Fako, 1985). This pattern is related both directly to labor migration and to late marriage. Traditionally prohibitions against premarital sex and pregnancy were very strong (Schapera, 1933). By the 1930s they were already collapsing as Christianity, schooling, and labor migration disrupted the mechanisms whereby parents and the community maintained their authority over the young. Today young people are almost completely free of such control—often they provide the income that supports the older generation—and almost all of them have sexual relationships (Comaroff and Roberts, 1977). For many young women, especially those without jobs or who already have chidren, the presents and assistance they get from their boyfriends are an economic necessity (Molenaar, 1980).

The current fertility data for Lesotho reveal that divorced, separated, and widowed women also have high levels of fertility. Widowed women had 86 percent of the births in the year before the survey that they would have done if they were married. For divorced and separated women the equivalent figure is 73 percent. Given the small number of unmarried men in Lesotho, it is clear that many of these women have married lovers. Some of these relationships are a recognized form of concubinage known as bonyatsi (Murray, 1981). This is a long-term and open relationship in which the man provides a measure of support for his mistress. Such relationships are also common in Botswana. The 1930s village studies link the spread of concubinage to the decline of polygyny and emphasize that the woman had nearly always been married previously or was regarded as too old to marry (Schapera, 1940). In

Lesotho it is fairly acceptable for widowed women to conduct discreet affairs although any overt relationships may threaten their rights to the house-property complex established by their dead husband. Any children that they bear are considered the legal heirs of their dead husband (Gay, 1980). The somewhat lower fertility of divorced and separated women is probably related to their more insecure social and economic position. In jural terms they become reattached to their natal household and they usually lose all rights to the dwelling, land, and productive wealth of the "house" established by their marriage (Gay, 1980). Some deserted women may be able to remain in their marital home and the fathers or brothers of divorced and separated women may be in a position to allocate them some land. Even women with access to fields, however, are unlikely to have the cash resources needed to farm successfully. Moreover, there is still considerable stigma attached to returning to live in the natal home (Gay, 1980). As a result divorced and separated women are particularly likely to become migrants. To interpret their fertility patterns we need to consider the interrelationships between migration and childbearing for women.

Female Migration and Fertility

The extent to which the migration of women from Lesotho is related to their marital status and household circumstances can be seen from table 8.10. The absolute number of married female migrants is large, but married women are much less likely to migrate than single, divorced and separated women. Bearing in mind that not everyone who slept elsewhere the previous night is a migrant—or every migrant currently an absentee—it is clear that, for

women, migration is incompatible with maintaining their own household at their place of origin. Those married women who do migrate are almost all living with their parents-in-law and those unmarried women, mainly widows, who head households in the rural areas are highly unlikely to be migrants themselves. Female labor migration is concentrated among the minority of women whose marital status tends to exclude them from access to both agricultural land and remittances from male migrants (Gay, 1980). Writing about Botswana over 40 years ago, Schapera (1940) noted the tendency for unmarried mothers, divorcees and separated women to be pushed into migration in order to obtain support for their children. The migration of such women reflects their lack of alternative sources of income. In this respect the economic situation of divorced and separated women is similar to that of the small minority of single women with children in Lesotho and the much larger number of such women in Botswana. In the Basotho village studied by Gay (1980), 49 percent of single women and 54 percent of divorced and separated women had some form of wage employment compared with 8 percent of married women and 14 percent of widows. Such work as is available both in the formal and in the informal sectors, for example beer brewing, is largely insecure and poorly paid. For some women, moving to an urban area may be an attractive opportunity, but female migrants with children usually have to foster them with their mother or other relatives. They migrate because they are forced to as the only way of fulfilling their maternal responsibilities.

The links between female migration and fertility are neither simple nor unidirectional. Not only are unmarried women with children pushed into migration but women living in the townships of the Republic of South Africa are both less subject to the social pressure militating against premarital pregnancy and less likely to meet potential husbands than those who remain at home (Murray, 1981). In the short term, however, the links between female migration and fertility operate in the opposite direction. Migrants who become pregnant may lose their jobs and women breastfeeding young children are unable to take up work away from home (Izzard, 1985). In this light the fertility differentials shown in table 8.11 are hard to interpret. Clearly much of the reduction in the fertility of widowed and divorced women in Lesotho is associated with their involvement in labor migration. What is less clear is to what extent migrant women are restricting their fertility through contraception and abstinence and to what extent caring for an infant prevents migration.^[2] To us it seems highly likely that both factors are important.

Male Migration and Marital Fertility

The direct impact of spousal separation on the level of fertility within unions depends to a considerable extent on its timing relative to birth intervals (Hill, 1985; Millman and Potter, 1984). The LFS household data in table 8.12

show that in Lesotho women whose husbands were absent the previous night had higher fertility over the previous year than women whose husbands were present. To some extent this may be because migrants are drawn preponderantly from high fertility groups of the population. Probably more importantly, spells of labor migration often follow the birth of a child. In part this correlation arises simply because most births are conceived during periods of leave. It also seems likely that the birth of a child provides an extra economic incentive to migration. In addition Basotho and Tswana culture prescribe a period of separation of husband and wife after birth (the woman normally returns to her natal home for several months) and prolonged postpartum abstinence (Gay, 1980; Manyeneng et al., 1985). Thus migration in the period following birth is in accordance with local values.

Information on the frequency of spousal separation at times when the woman would otherwise be exposed to the risk of conception is available from the LFS. It enquired about absences of three or more months during the open birth interval whenever the respondent was married and not pregnant, sterilized, or abstaining after a birth. Analysts who have considered the retrospective data have judged them incomplete and difficult to interpret (Casterline et al., 1984; Mpiti and Kalule–Sabiti, 1985). However, if women

reported any absences, they were asked if their husband was still away. These current status data are probably more accurate and are easier to interpret. From them one can identify the number of exposed women whose husbands are absent. Using this figure one can calculate a prevalence-incidence mean duration of absence during the period of exposure in each birth interval.^[3] In all, 16.5 percent of otherwise exposed married women said that their husband was absent. This suggests that women's husbands are away on average for about 3.4 months between the end of the postpartum nonsusceptible period and when they become pregnant, lengthening birth intervals by an equivalent amount. This is very much a minimal estimate of the direct impact of labor circulation on martial fertility. The questions posed in the survey asked about absences of 3 or more months, but even short periods of separation will affect fertility if they occur on a wide enough scale. The 1978/1979 Labour Force and Migration Survey of Lesotho showed that 20 percent of currently absent miners had visited home two or more times in the last 6 months (UN Economic Commission for Africa, 1982). Almost none of them would be detected by the LFS questions although on average they had spent only 1 month at home in this period. Moreover, even if the average interval between periods of leave for the 80 percent of migrants who return home less often was as long as 9 months, a third of them would have been omitted from the LFS statistics. According to the LFS household survey, only 54 percent of married men aged less than 55 slept at home the previous night. Therefore, even though migration is concentrated in women's postpartum nonsusceptible period, it seems likely that the LFS questions picked up less than half the demographically significant episodes of spousal separation and that male migration adds at least 6 to 7 months to the length of birth intervals.

There is a second way in which spousal separation affects the length of birth intervals. Not only prolonged breastfeeding but also a prolonged period of postpartum abstinence is still normal in Botswana and Lesotho (Manyeneng et al., 1985; Mpiti and Kalule–Sabiti, 1985). Data from the Botswana Family Health Survey shown in table 8.13 demonstrate that women with absent husbands whose last birth occurred more than 6 months ago were much more likely to report that they were abstaining than women whose husbands were present. In Lesotho, women who were abstaining were not asked about spousal separation and part of the reduction in fertility attributed to abstinence could more reasonably be attributed to labor migration. The published data on Botswana do not permit one to make exact calculations but the mean duration of postpartum abstinence estimated from data on women whose husbands were present at the time of the survey is about 9 months, compared with an estimate of 12 months based on all women.^[4] In Lesotho the mean duration of postpartum abstinence reported was about 15 months (Mpiti and Kalule–Sabiti, 1985). As prolonged periods of spousal separation are much more common in Lesotho than in Botswana, it seems

likely that anything between 3 and 6 months of the lengthening of birth intervals ascribed to postpartum abstinence in Lesotho occurs only because of the separation of married couples resulting from labor circulation.

Modern Methods of Fertility Control

Several southern African countries are experiencing the initial phases of fertility transition. In the RSA the total fertility rate of the black population probably declined from around seven to just above five children during the 1960s and 1970s (Science Committee of the President's Council, 1983). A fertility survey conducted in 1982 showed that 45 percent of married women were currently using contraception (van Tonder, 1985).^[5] Surveys conducted in Botswana and Zimbabwe in 1984 showed that 28 percent and 38 percent respectively of women currently in a union were using contraception (Manyeneng et al., 1985; National Family Planning Council, 1985). Although the LFS revealed fewer users of contraception, the position could have changed greatly during the last decade and even in 1977 there was evidence of fertility decline among educated women (Timaeus and Balasubramanian, 1984). A preliminary examination of the provisional results of the recent censuses of Botswana and Zimbabwe suggests that, so far, the adoption of modern methods of contraception has had little impact on the overall level of fertility. The total fertility rate in both countries is still in the range 6.5–7.0 children (Central Statistics Office, 1972; Manyeneng et al., 1985; National Family Planning Council, 1985). Moreover the mean completed family size expected by women, although lower than in many sub-Saharan African countries, remains as high as six children (Manyeneng et al., 1985; National Family Planning Council, 1985). A similar figure was reported in the LFS (Lightbourne, 1985). As in the RSA, where black women still want 4.4 children on average (van Tonder, 1985), fertility decline may proceed more slowly than it has in other parts of the world.

It is impossible to quantify the extent to which the migrant labor system has encouraged the adoption of modern methods of contraception. None of the published reports on the fertility surveys conducted in Southern Africa investigates differentials in contraceptive use according to whether women or their partners are involved in labor migration. Furthermore none of them examine the influence of marital status on recent fertility or contraceptive use. Most of the tables are restricted to currently married women. Yet, despite the lack of statistical evidence, there are reasons for supposing that the pervasive influence of labor migration on southern African society extends to this development. We would not disagree with those, such as Cleland (1985), who see the onset of control of marital fertility as a cultural innovation. But, as Cleland implies, there are at least two reasons why this represents an incomplete explanation. First, other social changes may be important preconditions for the diffusion of the idea of family planning. The populations of Southern African countries have been involved in wage employment and migration to urban areas for several generations. Although female migration is not very prevalent, a substantial proportion of women spend some period of their life working in a town. Moreover, whether or not they work outside the home, the wives of migrants are accustomed to making decisions, managing their households, and controlling their own lives. Partly in response to the emergence of large numbers of female-headed households, the rights of women to own productive wealth and be independent of men are being recognized gradually (Gay, 1980; Molenaar, 1980; Poulter, 1976). In addition, levels of female education are high in Southern Africa. Thus the wide-ranging series of changes in the status of women associated with labor migration and the long and profound exposure of the indigenous population of Southern Africa to Western ideas and values have probably removed any institutional barriers to the adoption of new attitudes to family building. Second, we would argue that, as Cleland (1985) admits, there are strong cultural supports for high fertility in Africa which may be based on social structural factors. Cleland suggests that the importance of the numerical strength of kin groups may be relevant. In Botswana and Lesotho, the growth of wage earnings in combination with their distinctive land tenure systems mean that such considerations no longer apply. While political and economic incentives to maximizing fertility appear to have existed in the nineteenth century, they probably ceased to operate as long as 50 years ago. It may be, as we have already speculated, that despite this women in Southern Africa will continue to want large families. If most women are using contraception to control the timing and spacing of their births, rather than to reduce the size of their families, the overall impact on the level of fertility will be slight. On the other hand, the strong and immediate economic disincentives to childbearing for single, divorced, and separated women in Botswana and Lesotho suggest that even this pattern of contraceptive use could lead to a substantial decline in fertility outside marriage.

The Overall Level of Fertility

In order to assess the overall significance of labor circulation for fertility it is helpful to examine the impact of the more important proximate determinants of fertility on its level using the model developed by Bongaarts (Bongaarts, 1978; Bongaarts and Potter, 1983). The data required are only available for Lesotho. We would emphasize that such an analysis does not permit one to make counterfactual predictions. While the development of widespread migration and late marriage were associated, it does not follow that a reduction in the extent of migration would lead to earlier marriage. More generally, the impact of the total and permanent cessation of labor migration on Lesotho's population would be disastrous and the implications for fertility impossible to determine.

Estimates of the fertility-inhibiting effects of the proximate determinants of fertility are shown in table 8.14. Our model is a revised and extended version of the one presented by Mpiti and Kalule-Sabiti (1985). We prefer to base our estimates of overall fertility and marital fertility on the WFS household survey (Timaeus and Balasubramanian, 1984) and have included spousal separation in the model. The total fertility rate of 5.4 is extremely low for a sub-Saharan African country. Even our minimal estimate of the extent of spousal separation at times when women would otherwise be exposed to the risk of conception suggests that it reduces the level of marital fertility by 9 percent. If we assume that the LFS questions underestimated such separation by a factor of two and that abstinence arising from spousal separation lengthens the postpartum nonsusceptible period by 3 months, a 24 percent reduction in fertility results. In other words, a hypothetical country with

similar marriage and fertility patterns to Lesotho but no labor migration would have a total fertility rate of at least six children and more probably one of around seven children.

The low level of premarital fertility in Lesotho means that the high mean age at marriage of women exerts an appreciable downwards influence on the level of fertility. In contrast marital dissolution has a trivial impact on the total fertility rate. It is worth emphasizing that this is not because there are few widowed, divorced, and separated women or because they are concentrated at the upper end of the reproductive age range. It is because of the high level of extramarital fertility. If no births occurred outside marriage, the current pattern of marital dissolution would result in an 11 percent reduction in the level of fertility in Lesotho (Casterline et al., 1984). Given that female migration is concentrated among divorced and separated women, it follows from the unimportance of marital dissolution for fertility that labor migration of women has little effect on the national level of fertility.

Discussion

It is sometimes argued that because childbearing often occurs outside marriage in sub-Saharan African nuptiality is irrelevant to the analysis of fertility.^[6] We believe that this is a misconception for the following reasons. First, even when there are few sanctions against extramarital childbearing, the fertility of unmarried women is lower than that of those who are married. In Lesotho premarital pregnancy remains fairly uncommon despite high ages at marriage of women. Second, in Southern Africa at least, the increase in childbearing outside marriage is a fairly recent development. In Botswana and Lesotho extramarital fertility and the transformation of marriage patterns are two aspects of the same process of change. If we wish to explain why births occur outside marriage, we have to understand the marriage system (Comaroff and Roberts, 1977). Third, marital status has important implications for women's social and economic circumstances. It determines the recruitment of children to social groups, the allocation of responsibility for their upbringing, and rights to their labor as they grow older. In Botswana it seems likely that a group of elderly men who have never married and have no legitimate children will emerge, raising new issues concerning support of the old. The social significance of childbearing inside and outside marriage is different as are the implications for the welfare of children and even their mortality. As a result, adoption of modern methods of fertility control may have different implications for the fertility of married and unmarried women.

As well as having a direct impact on fertility, the growth of widespread labor migration in Botswana and Lesotho played a major part in the development of the contemporary marriage systems of these countries. At the individual level we have shown how, for both men and women, involvement

in and the timing of labor migration shape and are shaped by their marriage and fertility histories. Late marriage, the decline of polygyny, increased marital disruption, and frequent extramarital sexual relationships are a series of interrelated changes that have reinforced one another and are all linked to labor migration. Other factors, in particular the spread of education, especially of women, the influence of the Christian church, and wider aspects of the development of a money economy and of the emancipation of women, are also relevant to the process of change. However the influence of labor migration is pivotal. It lies at the heart of a transformation of the role marriage plays in the social system and, more generally, of relationships between men and women and successive generations.

In traditional society, which we would loosely locate in the nineteenth century, women's lives were dominated by their key contribution to subsistence agriculture and their subordination to men (Izzard, 1981). Their participation in the political and ritual life of the community was restricted and they could neither own nor control productive wealth. For men marriage was an avenue to the control of the economic and political resources represented by women's productive and reproductive capacities. The payment of bridewealth transferred these irrevocably to the husband's family. Polygynous marriage, the accumulation of cattle, and the establishment of relationships of clientage were related practices that could be used to amass wealth and power. The monopoly over economic resources held by senior men lay at the root of their control over both women and the younger generation. We would argue from the anthropological record that maximizing fertility, and therefore for men their number of wives, was economically advantageous. This stemmed at least as much from the speculative dynamics of the political, land-tenurial, and technological context as from the net wealth flow obtained directly from the labor of children (Caldwell, 1982).

Labor migration disrupted traditional relationships of authority both directly, by removing elders and young men from their households, and by providing new avenues for gaining a livelihood and accumulating wealth (Murray, 1977). As the importance of wage earnings grew, other political and cultural changes initiated by contact with Europeans undermined the viability of traditional means of amassing resources. As a result marriage lost much of its economic and political rationale for men while at the same time sanctions against sexual relationships outside marriage weakened and conjugal relationships came under increasing stress as many men began to spend most of their young adult life away from their permanent home. The defining characteristic of marriage remains the exchange of bridewealth for social paternity. A decline in the significance of rights in the younger generation underlies both the transformation of marriage patterns and the increase in other forms of sexual relationship. The apparent continuity in the structure of marital institutions, which remain shaped by Basotho and Tswana cultu-

ral premises, is belied by a change in their purpose. To quote Murray (1981) again:

it is often more realistic in contemporary practice to represent marital transactions as the result of bargaining conducted by senior women over the earning capacity of men, than as the result of bargaining conducted by senior men over the productive and reproductive capacities of women.

In both countries most men's long-term security still depends on investment in the rural economy. They desire heirs and marry eventually. The earnings and contribution to domestic work of the younger generation are an important source of support in old age (Murray, 1977). But, in Botswana, ages at marriage are dispersed, reflecting the diversity of economic options available to different groups of men and women. For migrants there is little incentive to marry early. In contrast, in Lesotho, involvement in migration is more wide-spread and marriage patterns are more uniform. Marriage occurs at younger ages as only married men can begin to establish viable farming enterprises to provide for their retirement. There is nothing inevitable about these developments in the marriage system. As well as being shaped by the systematic exclusion of women from work in the RSA, they are contingent on the cultural and legal premises of Basotho and Tswana society. In particular, they are underpinned by differentiation between male and female labor, men's privileged access to land and cattle, and the identification of marriage with the transfer of bridewealth (Kuper, 1985).

The development of the migrant labor system created new roles and responsibilities for the wives of migrants and for migrant women. To a considerable extent these have become accepted and institutionalized both culturally and legally. This and related developments, such as the spread of female education, have greatly enhanced women's status. Nevertheless, their responsibility for child rearing, restricted access to land, exclusion from inheritance, and lack of opportunities for paid employment mean that most women, whether married or not, remain dependent on relationships with men to obtain the means of survival. Kraeger (1982) has suggested that:

demographic facts may in important respects constitute groups in some of their most distinctive features. In characterizing different demographic regimes we should ask: To what social purposes are demographic relations organized?

What we would emphasize is that the social purposes underlying demographic patterns may vary between groups of the population that differ in their economic situation and social status. A reproductive regime represents a negotiated or imposed resolution of such conflicts. In both countries a central concern of men in marriage and reproduction is to establish their rights in a farm and the next generation in order to provide for their retirement. For women in Lesotho and also many women in Botswana, it is crucial to gain

and maintain access to economic resources controlled by men. For the parental generation, the marriage of their daughters is an opportunity to obtain bridewealth cattle from younger men. Prolonged spousal separation adds to the stresses imposed on the conjugal relationship by these conflicting interests. Whether it survives or collapses, it is often women with young children who lose the most.

The differences between the marriage systems of Botswana and Lesotho are a telling caution against assuming that our analysis can be extended in every detail to other Southern African countries. On the other hand, many migrants in such countries also retain links with their village of origin despite long periods of absence. Moreover patrilineal kinship systems and marriage with bridewealth are found throughout most of the region to the south of Zambia and Malawi, although the focusing of property rights on a woman's "house" is confined to the area south of the Limpopo River. Tentatively we would suggest that Lesotho may typify areas where it remains important for male migrants to retain a link to a rural farming enterprise whereas elsewhere, including urban areas, the disruption of family life will be even greater. The data that are available about marriage and fertility patterns in other Southern African countries are consistent with our expectations. In the RSA the mean age at marriage of black women is above 20 years and polygynous unions are rare and had already become uncommon by the 1940s (see chapter 6 and Clignet, 1970). According to the fertility survey conducted in 1982, 44 percent of both Sotho and Nguni women had borne a child before they were married (van Tonder, 1985). Anthropological evidence also attests to the disruption of marital institutions and high level of fertility outside marriage in the Republic of South Africa (Dubb, 1974; Pauw, 1973) Similarly in Zimbabwe only 27 percent of 15 to 19-year-old women are married but 11 percent of all women of reproductive age are living in informal unions (National Family Planning Council, 1985). Thus there is some wider evidence that the complement of the distinctive Southern African marriage pattern identified by Lesthaeghe and his colleagues (see chapter 6) consists of other forms of sexual relationship and extramarital fertility.

Our attempts to measure the direct impact of labor migration on fertility have been bedeviled by lack of data, inadequacies in the questions used to investigate the subject, and difficulties in establishing the direction of causality. In Lesotho, the only country for which we could make estimates, almost every household is involved in migration. It is therefore impossible to measure any but the most short-term differentials between the fertility of migrants and their spouses and that of nonmigrants. Nevertheless we have firm evidence that male migration reduces the level of marital fertility by at least 9 percent and some grounds for suggesting that the actual reduction in fertility is very large, perhaps in the region of 25 percent. In Botswana patterns of breastfeeding, postpartum amenorrhea, and abstinence, on the one hand,

and of migration, on the other, are broadly comparable with those in Lesotho. The proportion of the male population involved in labor migration is about half of that in Lesotho. While we would not attempt to estimate a figure, it seems likely that spousal separation has an appreciable impact on fertility. In contrast, although migrant women have a low level of fertility, such migration is not common enough in Lesotho for it to have any significance for the overall level of fertility. Effects operating in the opposite direction are very important. Women's marital and reproductive histories are major determinants of their direct involvement in the migrant labor system.

Use of modern methods of contraception is now common in at least Botswana, the RSA, and Zimbabwe. This contrasts with the slow adoption of contraception in other sub-Saharan African countries including those such as Ghana and Kenya with active family planning campaigns. One can only speculate as to whether labor migration is the crucial feature distinguishing Southern Africa from the rest of the continent in this respect. Certainly the political and economic history that has transformed the indigenous population of South Africa and its periphery into an impoverished rural proletariat has destroyed many potential obstacles to the spread of the idea of controlling fertility. It remains to be seen whether a rapid decline in fertility is resulting, as it has elsewhere in the world, or whether women will continue to have large families. One theme of our argument has been that changing economic opportunities and social disruption resulting from labor migration led to the replacement of a reproductive regime in which marriage for women was early and universal by one in which a variety of forms of sexual relationship without payment of bridewealth are common. The result was a high level of fertility outside marriage. Use of modern methods of contraception makes it possible for unmarried women to be sexually active without having children. If for no other reason, the desperate economic plight of many such women may be leading them to adopt this course.

Acknowledgments

We would like to thank Allan Hill and Colin Murray for their comments on an earlier version of this paper.

Bibliography

Allison, C. 1979. The determinants of primary schooling in the Kweneng with special reference to cattle and mine labour migration. Working Paper of the National Migration Study. Gaborone, Botswana: Central Statistics Office.

Ashton, C. H. 1967. The Basotho, 2d ed. Oxford: Oxford University Press.

Bell, M. 1986. Contemporary Africa. London: Longman.

Bongaarts, J. 1978. A framework for analyzing the proximate determinants of fertility. Population and Development Review 4(1): 105–132.

Bongaarts, J., and R. Potter. 1983. Fertility, biology and behavior: An analysis of the proximate determinants. New York: Academic Press.

Boserup, E. 1970. Women's role in economic development. London: Allen and Unwin.

Bozzoli, B. 1983. Marxism, feminism and South Africa studies. Journal of Southern African Studies 9(2): 139–171.

Brown, B. 1979. Women's role in development in Kgatleng District. Working Paper of the African Studies Center. Boston: Boston University.

Bryant, C. 1977. Women migrants, urbanization and social change: An African case. Paper presented at the meeting of the American Political Science Association, Washington, D. C.

Bundy, C. 1979. The rise and fall of the South African peasantry. London: Heinemann.

Caldwell, J. C. 1982. Theory offertility decline. London: Academic Press.

Casterline, J. B., S. Singh, J. Cleland, and H. Ashurst. 1984. The proximate determinants of fertility. World Fertility Survey Comparative Studies, no. 39. Voorburg: International Statistical Institute.

Chaney, E. 1980. Women in international migration: Issues in development planning. Washington, D. C.: U.S. Agency for International Development.

Clarke, D. C. 1977. Foreign migrant labour in southern Africa. Geneva: International Labour Organization.

Cleland, J. 1985. Marital fertility decline in developing countries: Theories and evidence. In: Reproductive change in developing countries, ed. J. Cleland and J. Hobcraft. Oxford: Oxford University Press.

Cliffe, L. 1977. Labour migration and peasant differentiation: Zambian experiences. Journal of Peasant Studies 5: 326–346.

Clignet, R. 1970. Many wives, many powers—Autonomy and power in polygynous families. Evanston: Northwestern University Press.

Cock, J. 1980. Maids and madams. Johannesburg, South Africa: Raven Press.

Colclough, C. 1980. Some aspects of labour use in southern Africa—Problems and policies. IDS Bulletin 11(4): 29–39.

Colclough, C., and S. McCarthy. 1980. The political economy of Botswana: A study of growth and distribution. Oxford: Oxford University Press.

Comaroff, J. L., and S. Roberts. 1977. Marriage and extra-marital sexuality: The dialectics of legal change among the Kgatla. Journal of African Law 21(1): 97–123.

Cooper, D. C. 1979. Economy and society in Botswana: Some basic national socioeconomic coordinates relevant to an interpretation of the National Migration Study statistics. Working Paper of the National Migration Study. Gaborone, Botswana: Central Statistics Office.

Davies, R. 1978. Demand trends for foreign workers and employment strategy in South Africa. Paper presented at the Economic Commission for Africa, Conference on Migratory Labour in Southern Africa, Lusaka, Zambia.

Doyal, L. 1979. The political economy of health. London: Pluto Press.

Dubb, A. A. 1974. The impact of the city. In: The Bantu-speaking peoples of southern Africa, ed. W. D. Hammond-Tooke. London: Routledge and Kegan Paul.

Fako, T. T. 1985. Youth education and services for health and family life. London: International Planned Parenthood Federation.

Gay, J. 1980. Basotho women's options: A study of marital careers in rural Lesotho. Ph.D. dissertation. Cambridge University.

Gluckman, M. 1971. Marriage payments and social structure among the Lozi and Zulu. In Kinship, ed. J. Goody. London: Penguin.

Gordon, E. n.d. Towards a final report: The women left behind—The wives of migrant workers of Lesotho. Geneva: International Labour Organization.

Gulbrandsen, O. 1986. To marry—or not to marry. Ethnos 51(1–2): 7–28.

Hill, A. G. 1985. A practical guide to estimating the Bongaarts indices of the proximate determinants of fertility. In Population factors in development planning in the Middle East, ed. F. C. Shorter and H. Zurayk. New York: The Population Council.

Izzard, W. J. 1981. The impact of migration on the roles of women. Paper presented at the National Migration Study Conference, Gaborone, Botswana.

———. 1982. Rural-urban migration in a developing country: The case of women migrants in Botswana. Ph.D. dissertation. Oxford University.

———. 1985. Migrants and mothers: Case studies from Botswana. Journal of African Studies 11(2): 258–280.

Kerven, C. 1979a . Botswana mine labour migration to South Africa. Working Paper of the National Migration Study. Gaborone: Central Statistics Office.

———. 1979b . Urban and rural female-headed households, dependence on agriculture. Working Paper of the National Migration Study. Gaborone: Central Statistics Office.

———. 1982. The effects of migration on agricultural production. In Migration in Botswana: Patterns, causes and consequences, ed. C. Kerven. Gaborone: The Government Printer.

Kingdom of Swaziland. n.d. Report on the 1976 Swaziland population census. Mbabane, Swaziland: Central Statistical Office.

Kooijman, K. 1978. A report on the village of Bokaa. Gaborone, Botswana: Botswana Extension College.

Kowet, D. K. 1978. Land, labour migration and politics in southern Africa: Botswana, Lesotho and Swaziland. Uppsala: Scandinavian Institute of African Studies.

Kraeger, P. 1982. Demography in situ. Population and Development Review 8(2): 237–266.

Kuper, A. 1985. African marriage in an impinging world—The case of southern Africa, Leyden: Afrika Studiecentrum.

Leggasick, M., and F. Le Clerq. 1978. The origins and nature of the migrant labour system in southern Africa. Paper presented at the Economic Commission for Africa, Conference on Migratory Labour in Southern Africa, Lusaka, Zambia.

Lesetedi, D. 1984. Education—1981 population and housing census. Gaborone: Central Statistics Office.

Lightbourne, R. E. 1985. Individual preferences and fertility behaviour. In Reproductive change in developing countries, ed. J. Cleland and J. Hobcraft. Oxford: Oxford University Press.

Lucas, B. 1982. Outmigration, remittances and investment in rural areas. In Migration in Botswana: Patterns, causes and consequences, ed. C. Kerven. Gaborone: The Government Printer.

McDonald, P. 1985: Social organization and nuptiality in developing societies. In Reproductive change in developing countries, ed. J. Cleland and J. Hobcraft. Oxford: Oxford University Press.

Magubane, B. 1975. The "native reserves" (Bantustans) and the role of the migrant labour system in the political economy of South Africa. In Migration and development: Implications for ethnic identity and political conflict, ed. H. I. Safa and B. M. Dutoit. The Hague: Mouton.

Manyeneng, W. G., P. Khulumani, M. K. Larson, and A. A. Way. 1985. Botswana family health survey, 1984. Columbia, Md.: Westinghouse Public Applied Systems.

Marks, S., and E. Unterhalter. 1978. Women and the migrant labour system in southern Africa. Paper presented at the Economic Commission for Africa, Conference on Migratory Labour in Southern Africa, Lusaka, Zambia.

Meillassoux, C. 1975. Maidens, meal and money. Cambridge: Cambridge University Press.

Millman, S. R., and R. G. Potter. 1984. The fertility impact of spousal separation. Studies in Family Planning 15(3): 121–126.

Mitchell, J. C. 1969. Structural plurality, urbanization and labour circulation in southern Rhodesia. In Migration: Sociological studies, ed. J. A. Jackson. Cambridge: Cambridge University Press.

———. 1985. Towards a situational sociology of wage-labour circulation. In Circulation in third world countries, ed. R. M. Prothero and M. Chapman. London: Routledge and Kegan Paul.

Molenaar, M. 1980. Social change within a traditional pattern: A case study of a Tswana ward. Leyden, Netherlands: Institute of Cultural and Social Studies.

Mosley, W. H., L. Werner, and S. Becker. 1982. The dynamics of birth-spacing and marital fertility in Kenya. World Fertility Survey Scientific Reports, no. 30. Voorburg: International Statistical Institute.

Mpiti, A. M., and I. Kalule-Sabiti. 1985. The proximate determinants of fertility in Lesotho. World Fertility Survey Scientific Reports, no. 53. Voorburg: International Statistical Institute.

Murray, C. 1977. High bridewealth, migrant labour, and the position of women in Lesotho. Journal of African Law 21 (1): 79–96.

———. 1981. Families divided. Cambridge: Cambridge University Press.

National Family Planning Council, Zimbabwe. 1985. Zimbabwe reproductive health survey 1984. Columbia, Md.: Westinghouse Public Applied Systems.

Palmer, R., and N. Parsons. 1977. The roots of rural poverty in central and southern Africa. London: Heinemann.

Parsons, N. 1977. The economic history of Khama's country in Botswana, 1844–1930. In The roots of rural poverty in central and southern Africa, ed. R. Palmer and N. Parsons. London: Heinemann.

Pauw, B. A. 1973. The second generation, 2d. ed. Oxford: Oxford University Press.

Poulter, S. 1976. Family law and litigation in Basotho society. Oxford: Clarendon Press.

Preston-Whyte, E. 1974. Kinship and marriage. In: The Bantu-speaking peoples of southern Africa, ed. W. D. Hammond-Tooke. London: Routledge and Kegan Paul.

Prothero, R. M., and M. Chapman, eds. 1985. Circulation in third world countries. London: Routledge and Kegan Paul.

Republic of Lesotho. 1981. Lesotho fertility survey 1977, first report. 2 vols. Maseru, Lesotho: Bureau of Statistics.

Schapera, I. 1933. Premarital pregnancy and native opinion: A note on social change. Africa 45: 258–279.

———. 1940. Married life in an African tribe. London: Pelican.

———. 1947. Migrant labour and tribal life: A study of conditions in the Bechuanaland Protectorate. Oxford: Oxford University Press.

Science Committee of the President's Council, Republic of South Africa. 1983. Report on demographic trends in South Africa. Cape Town: The Government Printer.

Skinner, E. P. 1985. Labour migration and national development in southern Africa. In African migration and national development, ed. R. Lindsay. University Park: Pennsylvania State University Press.

Taylor, J. 1982. Changing patterns of labour supply to the South African gold mines, Tijdschrift voor Economische and Sociale Geografie 73(4): 213–220.

Timaeus, I. 1984. Mortality in Lesotho: A study of levels, trends and differentials based on retrospective survey data. World Fertility Survey Scientific Reports, no. 59. Voorburg: International Statistical Institute.

Timaeus, I., and K. Balasubramanian. 1984. Evaluation of the Lesotho fertility survey 1977. World Fertility Survey Scientific Reports, no. 58. Voorburg: International Statistical Institute.

United Nations Economic Commission for Africa. 1982. Lesotho—A demographic, socio-economic profile. Paper presented at the seminar on Population Planning for Development, Maseru, Lesotho.

van Binsbergen, W. M. J., and H. Meilink, eds. 1978. Migration and the transformation of modern African society. Leyden: Afrika Studiecentrum.

van der Wiel, A. C. A. 1977. Migratory wage labour: Its role in the economy of Lesotho. Lesotho: Mazenod Book Centre.

van de Walle, E. 1985. Community-level variables and institutional factors in the Study of African nuptiality. In The collection and analysis of community data, ed. J. B. Casterline. Voorburg: International Statistical Institute.

van Onselen, C. 1976. Chibaro: African mine labour in southern Rhodesia—1900–1933. London: Pluto Press.

van Tonder, J. L. 1985. Fertility survey 1982: Data concerning the black population of South Africa. Pretoria: Human Sciences Research Council.

Wilson, F. 1972. Migrant labour in South Africa. Johannesburg: South African Council of Churches.

Wolpe, H. 1972. Capitalism and cheap labour power in South Africa: From segregation to apartheid. Economy and Society 1: 425–456.

Wylie, D. 1981. Migration to freehold farms and to farms in South Africa. Paper presented at the National Migration Study Conference, Gaborone, Botswana.