3. Kinship and the Yoruba town

This chapter deals with the linked questions of settlement pattern and kinship organisation. It touches on two of the most important debates in the Yoruba literature of recent years. First, there are the origins and nature of Yoruba urbanism. The Yoruba are usually thought to be unique in tropical Africa in the number and size of their towns. According to the 1953 census, 22 per cent of the Yoruba population lived in towns of over 100,000 and 53 per cent in towns of over 5000 inhabitants (Bascom, 1959: 32). This poses two questions. First, why did this apparently unique settlement pattern evolve? Was it due to population density, trade, religion, political centralisation, or the wars of the 19th century? Second, can Yoruba towns in fact be called 'urban' at all? (Krapt-Askari, 1969; Schwab, 1965; Lloyd, 1973). The early urban sociologists, and particularly those of the Chicago school, had generally associated urban life with increased social mobility, a complex division of labour, and the breakdown of primary relationships. Paradoxically, a large proportion of Yoruba townsmen appeared to be farmers, while residence appeared to be organised on the basis of patrilineal descent. Faced with this, some writers have preferred to compare Yoruba towns with 'pre-industrial cities' in other parts of the world (e.g. Wheatley, 1970) or tried to show that they do indeed qualify for 'urban' status in terms of Wirth's well-known criteria of size, density, heterogeneity and permanence (Bascom,1960b; cf. Wirth, 1938).

Second, there is the nature of the Yoruba kinship system. Given the preoccupation of African anthropology with segmentary lineages in the 1940s and 1950s, it is hardly surprising that writers dealing with the Yoruba should have taken corporate descent groups as their units of analysis (Lloyd, 1955; Schwab, 1955). In 1962, Lloyd's study of Yoruba land law provided a detailed comparison of the social structures of four kingdoms which showed clearly the wide range of variation in social organisation in different areas of Yorubaland. In this study and in a later paper (1966b) he argued that the kinship system of the northern Yoruba kingdoms (including Oyo, Egba and Ekiti) were agnatic, while those of the southern kingdoms of Ijebu and Ondo were cognatic, though there is also a range of variation within each of these categories. He went on to relate this basic distinction to other differences: in land-tenure and settlement patterns, political systems (1965), divorce rates (1968a) and the degree of incorporation of women in their husbands' descent groups on marriage. In 1970, Bender challenged the classification of the Ondo system as cognatic, and pointed to the many features it shared with the agnatic systems of the kingdoms to the north. Lloyd responded with a brief but useful rejoinder which summarised the differences in their positions (1970).

In this chapter I attempt a reassessment both of Yoruba urbanism and of kinship patterns. The main arguments are these. Far from being unique, the process of development of Yoruba towns before the 19th century was similar to that which took place in a number of other West African states. Yoruba urbanism only assumed its characteristic form as a result of the 19th-century wars, and it was these historical factors which have produced the diversity of settlement patterns rather than differences in kinship organisation as Lloyd suggests. The debate about whether Yoruba towns are 'urban' has been superseded by another question: that of the differences between the economic and administrative roles of individual towns and their resulting growth patterns. With regard to kinship I argue that the division of Yoruba systems into agnatic and cognatic is artificial as there are important cognatic elements in the kinship organisation throughout Yorubaland. The relative strength of these elements in particular areas is again the outcome of historical factors, and the situation in some areas is still one of rapid change. Corporate patrilineal descent groups have not developed in all areas, even in the 'agnatic' parts of Yorubaland, and where they do exist, their development is a historical problem. They cannot simply be taken as given and their existence then used to account for other features of social organisation.

The Yoruba town

The Yoruba themselves commonly distinguish between the 'town', ilu, and 'farm villages', aba or abule. Ilu denotes a permanent settlement with its own government. Farm villages, on the other hand, are more temporary settlements where people stay if they are working on farms outside easy commuting distance, say an hour's walk away from the town. The distinction is irrespective of size. Some ilu have been in demographic decline for many years and are now very small, while some of the older-established farm villages, particularly in the cocoa-producing areas, have increasingly large permanent populations.

There is no such thing as a 'typical' Yoruba town, as far as spatial organisation is concerned, though many features are widely shared. In the 19th century many towns were heavily fortified and the outline of the walls is still occasionally visible. The gates were often controlled by the ruler or one of the subordinate chiefs for whom the tolls charged were an important source of revenue. Another source of revenue was the market which is often located in the centre of the town, next to the ruler's palace. Market revenues are now collected by local government. In smaller towns the markets consist of open spaces with, at most, a few temporary grass shelters. In the larger towns there are more permanent corrugated-iron or concrete stalls, put up by the local authority or individual traders. Lagos and Ibadan have numerous markets, the largest containing several thousand traders. Some of these act as regional centres for a particular trade, such as that in foodstuffs, cloth or kola nuts (cf. Hodder, 1967).
The palaces of Yoruba rulers are large, sprawling structures, consisting of a large number of courtyards, off which lead the public rooms, shrines and the private apartments of the ruler together with his wives and palace officials (Ojo, 1966a). In the 1930s the palace of the Alafin at Oyo consisted of thirty courtyards, kara, spread over 17 acres. Its predecessor at Oyo-lle may have extended over an area of a square mile. Some palaces still have within their walls an area of bush, once used for the collection of plants used in ritual as well as more mundane sanitary purposes. During the colonial period many rulers built European-style two-storey residences for themselves, while other parts of the palace were demolished to make way for district offices or a town hall.

The remainder of the town is divided for administration into a number of wards known variously as adugbo (Oyo and Ife), itun (Ijebu) and idimi (Ondo), each under the jurisdiction of one of the title-holders. In towns such as Ife and Ilesa, the wards extend out from the centre of the town, clustered around the main roads. Extending beyond these are the farms belonging to the descent groups within the ward. In the case of Ijebu Ode the farmland is limited, and the pattern of ward boundaries is more like a chess board (KrapfAskari, 1969: 51ñ2).

This pattern has been modified, particularly in the larger and more prosperous towns. Large areas are now given over to administrative buildings, government residential areas, commercial and industrial sites, and newer houses built specifically for rental by civil servants and local entrepreneurs on what was formerly farmland at the edge of the town. This is the pattern even in relatively undeveloped towns like Ogbomoso. Here there is a single ribbon of development skirting the town along the main road from Oyo to Ilorin. It contains most of the schools, both of the hospitals, the Baptist seminary, the premises of the banks and other commercial firms, and the shoe factory. In Ibadan the commercial area is much larger and has grown up around the railway station to the north-west of the older parts of the city.

The wards of the town are subdivided into compounds, ile, which may themselves be divided into a number of more or less separate houses. The relationship between residence and kinship will be considered in a moment. In the northern towns where I did my own work, the compounds consisted of a number of buildings, sometimes crowded together and surrounded with an outside wall, as in the older and more congested parts of the town, or laid out in a more regular grid pattern with the houses separated by alleyways. This grid pattern is apparently typical of southern Yoruba towns like Ijebu Ode. The word ile can refer either to the whole compound or a single house within it. The term agbo ile ('flock of houses') is sometimes used to indicate the whole compound, but usually ile suffices and the meaning is clear from the context.

It is sometimes stated that compounds in the northern Yoruba towns are rather larger than those in the southern areas. This is something of an oversimplification. It is true that some compounds in the northern towns are very large with several hundred residents, but there is also a large degree of variation. In Oje in Ibadan, for instance, compound populations ranged from 26 to 346 (B. Lloyd, 1967: 67) while in Osogbo they ranged from 15 to 450 (Schwab, 1965). The range in terms of actual numbers of buildings is equally great. In Igbeti in 1970 the smallest compound consisted of a single building, while the largest contained over thirty. The largest compounds are often those of the major chiefs, but compound sizes and fission are also a function of space. Those compounds with space in which they can expand remain large. Those in the more congested areas of the town divide when their populations get too large.

Buildings throughout Yorubaland are rectangular, with or without a central courtyard. In northern towns like Igbeti, many compounds used to consist of a single building arranged round a very large courtyard, with perhaps thirty or more rooms. Houses of this type were built up to the 1930s. Since 1950 most of them have been demolished or modified and smaller houses have replaced them. In some Ogbomoso compounds the newer houses have been built in the courtyards of the older ones.

More recent houses in this area are of two types. In the first, up to ten rooms are arranged round a central courtyard, leading to it through a passage or pd,ed,e. In the second, a number of rooms are arranged on each side of a central corridor, with a yard at the back of the house (cf. Crooke, 1966). This second type is common throughout Yorubaland.

The majority of houses are still built of mud, with corrugated-iron or aluminium roofs. Thatched roofs are still the norm in farm villages, but are now quite rare in the towns. Rooms can easily be demolished or added on where necessary. In older compounds the rooms are often poorly ventilated and badly lit, with small doors and no windows, but they are used mainly for sleeping and storage. Most social life goes on in the odede or the courtyard. In the more recent buildings, doors and windows are larger. Other improvements are possible if the owner has the money. Mud walls can be made more durable by rendering them with cement, and more durable and expensive still are houses constructed of cement blocks, with sub-ceilings in all the rooms. The really wealthy invest in ile petesi, 'upstairs' houses, costing thousands of naira. The proportion of ile petesi in a town is a good index of its wealth. In Ogbomoso perhaps a fifth of the houses were of this type, many of them built by migrants to Ghana and northern Nigeria. In 1970 in Igbeti there were just five. Many of the houses put up by migrants may stay virtually unoccupied for years. There are also startling contrasts. A two-storey cement house may form part of the same compound as houses built out of unrendered mud with thatched roofs. Life-style and furnishings can offer similar contrasts under the same roof. An educated man with a salaried job will probably try to furnish his room with a spring bed and mattress, foam-cushioned armchairs, a fan and radiogram (where there is electricity), and even a refrigerator. Meanwhile his senior relatives make do with sleeping-mats, a small eating-table or two, roughly made wooden chairs and a few storage boxes.

The allocation of rooms within a house is fairly standard. The builder or the most senior man usually has the best rooms, either near the main entrance or opening onto the balcony on the first floor in the case of a storey building. He will probably occupy a 'parlour' where he entertains, as well as a bedroom and a room for each of his wives. More junior men with a single wife are allocated a single room each. There may be a couple of rooms set aside for the teenage boys and girls to sleep in: younger children sleep with their mothers, on mats in the odede, or even in the courtyard during the hotter part of the dry season. Empty rooms can be used for storage or let out to tenants. If the building faces onto a street, the rooms in front may be converted into stores and let out to traders.

A variety of other amenities are found. Wealthier house-owners may install running water inside the house. The majority of houses in Ogbomoso relied on a single tap in the courtyard, and each woman had a storage pot of her own. Water closets are common in some parts of Ibadan and Lagos, but rarer elsewhere. In 1970, the better Ogbomoso houses still relied on bucket latrines emptied each night by the 'night soil' men. Igbeti and Igboho by contrast had no running water or sanitation at all. People relied on compound wells and the surrounding bush. In the dry season of 1970ñ1, the Igboho women were having to bring water from rivers between three and five miles away. There was no electricity either. Not surprisingly, the provision of water and electricity are key political issues in many areas.

The rise of Yoruba urbanism

Accounts of Yoruba urbanism which stress its uniqueness usually quote two sources: the 1953 census figures, and the eye-witness accounts of Europeans who travelled in the alea in the 19th century (e.g. Bascom, 1959; 1973; Mabogunje, 1968; Krapf-Askari, 1969). What is not so often made clear is that the impressive scale of urban development which they observed in most cases does not predate 1800. Some large urban centres like Ife, Ijebu Ode and Oyolle undoubtedly did exist long before this ñ Ijebu Ode is mentioned in Portuguese sources dating from the early 16th century ñ but they were the capitals of the largest Yoruba kingdoms of the period, and in their size they were unusual among Yoruba towns. The capitals of Benin, Asante, Dahomey and the Hausa states were of a similar order of size. Until 1800, therefore, it is likely that the Yoruba pattern of urbanism was not very distinctive. The reasons for the initial development of Yoruba towns were probably similar to those elsewhere in West Africa: a complex division of labour, craft specialisation giving rise to (and intensified by) long-distance trade, and the need to control the major trade routes. The Yoruba kingdoms lie on the boundary between the savanna and the forest and trade between the two zones must have developed at an early date. Oyo was well placed for access to the important trade route from Gonja across the Niger to the Hausa states. Theories that Yoruba towns developed primarily as ritual centres (Wheatley,1970) or as administrative centres for an incoming Ife, dynasty (Mabogunje, 1968: 76) must be regarded as speculative.

The changes that took place in the 19th century are immediately apparent from the 1953 census figures. Twelve Yoruba towns were enumerated with populations of over 40,000 (Bascom, 1959: 31), and these included nine of the ten largest towns in Nigeria. Eleven of them ñ Ibadan, Ogbomoso, Osogbo, Ife, Iwo, Abeokuta, Oyo, Ilesa, Iseyin, Ede, and Ilorin ñ were either founded during the 19th century or grew rapidly during it with the influx of refugees from the devastated areas around. The exception is Lagos, whose rapid growth in the 19th century depended initially on the slave trade, which also depended on the wars in the interior.

This suggests that, but for these wars, the pattern of Yoruba urbanism would have been far less distinctive than it was in 1953. Significantly, Ijebu, the Yoruba kingdom least disrupted by the wars, is also the one with the most decentralised settlement pattern. The settlement patterns in Ife, Egba and Oyo were all apparently much less centralised before 1800 than they are at present. The Upper Ogun area of Oyo was dotted with a large number of settlements which were later destroyed (Babayemi, 1971). The present towns are near refuge areas where the survivors congregated, often on hills which could easily be defended (Gleave, 1963). The destruction of the Egba towns and the resettlement of the survivors in Abeokuta is also well documented, and a similar process seems to have taken place, though on a smaller scale, around Ife, (Bascom, 1969a: 29).

The diversity of Yoruba settlement patterns probably results, in part, from this two-stage process of urban development. Lloyd (1962: 54ñ8) distinguishes three main types. In the first, exemplified by Oyo and Ado Ekiti, the capital of a kingdom is surrounded by its farmlands which extend to between three and ten miles from the town, and by farm villages. The people living in the villages claim membership of one or other of the compounds in the capital (cf. Goddard, 1965). Further away are the subordinate towns in the kingdoms with their own farms and villages arranged in a similar pattern.

In the second type, exemplified by Ibadan and Abeokuta, a very large capital is surrounded by farmlands and villages which extend up to 20 or 30 miles from the capital. Subordinate towns are fewer, and most of the farmers in the kingdom claim membership of a compound in the capital.

In the third type, which differs more markedly and which is exemplified by Ijebu Ode, the farmland surrounding the capital extends only for a mile or two. Beyond are many small villages, but these are permanent settlements and their residents do not claim membership of compounds in the capital. On the other hand, a rather smaller proportion of the residents of the capital are involved in farming than is the case in most kingdoms: most of the Ijebu farmers live in villages outside.

Why did these differences evolve? If we accept a two-stage developmental process for Yoruba towns, it implies that the Ijebu pattern represents an earlier settlement pattern which was much more widespread before the 19th century wars (cf. Law, 1977: 10). Lloyd attempts a sociological explanation and relates it to the difference between the agnatic and cognatic versions of Yoruba kinship. In the cognatic Ijebu system, he argues, a man could claim land from a number of different descent groups. If land was short in a village, individuals could move elsewhere, allowing village boundaries to remain static for a long period. In the agnatic systems, a man could claim land only from his father's descent group. As the descent groups of the major chiefs grew larger in size, through the acquisition of large numbers of wives and slaves, so their demand for land increased, resulting in the conquest or absorption of surrounding settlements.

A process like this may well have taken place in some instances, but there are problems with it as a general model. Firstly, it assumes that land is scarce, which is certainly not the case in all areas where these settlement patterns are found. Secondly, it takes as given the existence of agnatic descent groups as corporate land-holding units. Again, this is not universal. Indeed, it is possible in some areas that the land-holding functions of descent groups are a relatively recent development related to increasing population pressure and the scarcity of certain categories of land ñ for instance forest land suitable for cocoa.

It seems more plausible to relate the differences in settlement pattern to historical events. In Oyo, Ife and Egba, the 19th-century wars resulted in the destruction of large numbers of smaller settlements, and the concentration of the remaining population in a relatively few large towns. Since then there has been a gradual recolonisation of the surrounding rural areas, and the foundation of hamlets and villages whose residents still see themselves as members of descent groups in the towns. In Ijebu, on the other hand, the original pattern of dispersed rural settlement survived, and the majority of the farmers in the kingdom see themselves as permanent residents of the smaller villages. Non-agricultural occupations are concentrated in Ijebu Ode. In the cases of Ibadan and Abeokuta, the rural settlements have gradually become more permanent, and many are now commercial and administrative centres in their own right: their residents may still claim membership of a compound in the capital, but many visit it only infrequently. In some parts of Ibadan like Oje, apparently very few of the permanent residents are farmers, and the same is true of Abeokuta (B. Lloyd, 1967; P. Lloyd, 1962: 58). Perhaps the crucial test of Peter Lloyd's model, though, lies in the northern Ijebu towns of Ijebu Igbo and Ago Iwoye, which appear to have settlement patterns similar to those in the northern Yoruba kingdoms, rather than to that of Ijebu Ode (1962: 138). These lie near the border with Ibadan, with which Ijebu was at war at various times throughout the 19th century. Certainly Ago Iwoye was founded as a result of the destruction of other settlements (1962: 140). The evidence for Ijebu Igbo is less certain, but it is a plausible hypothesis that the reason for its size and settlement pattern was the need for defence in the 19th century, rather than its kinship system, which one assumes is similar to that elsewhere in Ijebu.

Nevertheless, even if Lloyd's model of the relationship between kinship and settlement pattern is not acceptable, that does not mean that there is no systematic relationship between the two, and an alternative approach will be suggested later in the chapter.

Kinship and social organisation

In earlier accounts of Yoruba social structure, writers such as Johnson (1921: 98ñ100) took as their starting point the compound: the large residential units into which the wards of most Yoruba towns are divided. Such an approach had much to commend it. Compounds were named, spatially distinct, entities, and the major social units with which the actors identified. They not only provided the framework for residential organisation, but they were important units of political organisation. In many areas the compounds of the major military leaders became extremely large in the 19th century, with an influx of followers and slaves, and so the compounds formed the core of political factions in the town.

Gradually there was a shift of emphasis in accounts of the Yoruba. Anthropologists working throughout Africa increasingly focused their attention on unilineal descent groups, and in the case of the Yoruba this resulted in the early work of Bascom (1944), Lloyd (1955) and Schwab (1955). Although they worked in different towns, their conclusions were similar. The population of the Yoruba town was seen as being divided into a number of corporate patrilineages. The male members of each typically reside together in a compound together with their wives and children. These lineages are exogamous units, with distinctive facial marks, food taboos and sets of names. They are also land-holding and title-holding units, and in some cases practise specialised occupations. They are usually segmented, though the process of segmentation is not regular, and may occur at any level in the genealogy.

Although this model is inadequate as the basis for a general discussion of Yoruba kinship, it must be said that it does approximate to reality in a number of northern Yoruba towns. Lloyd draws on a case study from Saki in Northern Oyo Division. In culture and social structure, Saki is very similar to Igbeti, and Figure 1 shows the relationship between residence and genealogy in an Igbeti compound.[1] Genealogies in Igbeti are short, extending back three generations beyond the present elders to the refoundation of the town in the 19th century. Genealogical memories throughout Yorubaland tend to be short, though they become longer when significant resources are at stake, such as valuable land controlled by the descent group.

In 1971 the compound consisted of nine houses and a mosque. The descent group which forms its core was founded by the great-grandfather of the present head or bale. In 1906, the British persuaded the population to move down from the top of an inselberg, where they had taken refuge during the wars, to the present site of the town. Originally this compound consisted of a single large building with a central courtyard. This was demolished in the 1950s, and three smaller houses were put up on the site. The other houses were added subsequently. These have used up the available land on this site, and anyone else wishing to build will have to move to another part of the town.

Of the houses which replaced the original building, one (No. 3 in the diagram) is occupied by the bale together with his senior wife, his senior son and his son's wives and children.

The second son of the bale by his first wife is a prominent Igbqti Muslim. He built his own house in the early 1950s (No. 4) and in 1973 he was converting it into an ile pStrsL He lives there with his four wives and children, but has since built a second house next to it (No. 5) where his senior son lives when he is in town. The other rooms are rented out to migrants. He was also responsible for building the mosque of which he is the Imam, and where he runs a koranic school.

The second wife of the bale now lives in a third house to the rear, built by her elder son (No. 6). He lives in Ghana with his two wives and children, but his junior brother, another koranic teacher, lives in it with his four wives and children.

The senior brother of the bale died in the 1940s. His two wives were still alive in 1970, and they occupied two more houses with their sons and families. The senior wife had three sons. The eldest son is dead, but his eldest son and family occupy his former room. The other rooms in this house (No. 1) are occupied by her other two sons and their wives and children. The two sons of the second wife live with their mother in a house to the rear (No. 2) built by the senior son, a trader in foodstuffs.

There are three other houses. One of them (No. 8) is occupied by the sons of half-brothers of the bale. One of them, now dead, built himself another house to the rear (No. 7) which is now rented to a local authority road foreman. His mother, who lives in No. 8, collects the rent. The last house in the compound (No. 9) was built by a more distant agnate who spent most of his life in his mother's natal compound, though he returned to his father's descent group to build. He lives with his junior brothers.

This example illustrates quite clearly some of the main patterns of segmentation in this area. Firstly, groups of siblings by the same mother, omoiya, tend to live together. Groups of siblings with the same father but different mothers may live together in their father's house at first, but they tend to separate when additional space is required. Segmentation is thus rooted in the polygynous family, and the often difficult relations between co-wives and their children. Trouble is most likely to arise between co-wives if the husband fails to give them equal treatment in some way, or if one wife is markedly more fertile than another. In order to appear impartial, the polygynous husband tends to adopt a rather distant relationship with all his wives (LeVine et al., 1967; cf. Awolowo, 1960: 21ñ2). Each wife has to look after the interests of her own children, and where this involves the provision of money, land or education, tension is more likely. On the other hand, the wives are precisely ranked in terms of seniority, with the senior one allocating work and organising the sleeping and cooking arrangements, and there is pressure from others in the compound for wives to resolve their differences. As a last resort, quarrelling wives can be moved to different houses, or even to different towns. The strains of the relationship are reflected in the evidence for higher divorce rates and even higher rates of mental disturbance in polygynous households (Lloyd, 1968a; Leighton et al., 1963).

Secondly, there is a tendency for an eldest son to stay in the same house as his father, and take over his room on his death. In Yoruba society, this relationship is very close, and an eldest son may stay in his father's house, even if his other brothers move elsewhere.

Thirdly, the process of segmentation is not at all regular. While some relatively deep segments remain together, others split up. The main factors are wealth and numbers. The wealthy men who usually have more wives and children are the most likely to build their own houses. Other factors which complicated the pattern in this case are evident. Normally full brothers collaborate in building, but here two of the sons of the bale decided to build separately. Both were wealthy and had large families, and the koranic teacher wanted to seclude his wives. This is common among Muslims in northern Nigeria, but very uncommon among the Yoruba. Personality factors also played a part. Relations between full brothers should ideally be close, but they are not always so. In another Igbeti compound, two brothers who had quarrelled over politics divided their house by building a wall in the middle of the courtyard and demolishing the adjacent rooms! This was regarded as rather an extreme case. The mother, if she is still alive, is often able to reconcile brothers who have differences, even if the quarrel is a really serious one, over land or money.

Other Igbeti examples, however, do not fit the model so well. Several Igbeti compounds which contain members of more than one descent group exist, while the largest compound in the towns has no less than six descent groups within it. Other 'complex compounds' of this type existed in the past, but the constituent descent groups have tended to separate, as a wealthy member of one of them builds his own house and takes his relatives with him, or as lack of space for further building leads to fission.

Thus the one-to-one relationship between compound and descent group is, in some instances, a recent development. Fadipe was well aware of these complex units (1970: 99), but with the later emphasis on patrilineages their significance has been overshadowed. Generally the Igbeti think of themselves as members of compounds rather than descent groups, and where, for instance, the descent compound contains more than one descent group, it is the compound rather than the descent group which is the exogamous unit.

A second point which Fadipe stressed, which tended to be overlooked with the stress on unilineal descent groups, was the important bilateral element in Yoruba kinship (1970: 134). As he says, bilateral kindred come into prominence at many points in the life history of the individual, during rites of passage, as well as in some questions of inheritance and succession. Of the later writers, Bascom (1944) did stress the importance of cognatic kinship, though he regarded it as subordinate to the patrilineal principle. Significantly, Schwab restricted his discussion of it to ego's relations with his mother's patrilineage. It was Lloyd who changed his position most radically. In 1955 he suggested that his model could be applied fairly generally to other Yoruba towns. By 1962 he had come to the conclusion that the southern kingdoms of Ijebu and Ondo in fact had cognatic kinship systems. In the case of Ondo this conclusion was challenged by Bender, and the next section takes this debate as a starting point for considering the whole question of cognatic elements in Yoruba kinship organisation.

Agnatic or cognatic?

In Yoruba Land Law, Lloyd described the social organisation of four Yoruba kingdoms: Ondo, Ijebu, Ado Ekiti and Egba. The kinship systems of Ekiti and Egba are discussed in terms of agnatic descent groups similar to those he described in his earlier paper. An individual belongs to only one descent group, that of his father, and membership is exclusive. These descent groups are exogamous corporate groups, and rights to property, land and titles are inherited within them. There are differences between the two kingdoms: in particular the corporate control of land by the descent groups in Ekiti has remained much stronger.

The Ijebu and Ondo systems on the other hand are based on cognatic descent groups: these groups are still defined with reference to a founding ancestor, but membership is open to his descendants through either male or female links. Thus membership is not exclusive, and an individual can maintain membership of more than one group simultaneously. In Ijebu and parts of Ondo, these groups form localised residential units, though most descent groups in the Ondo capital are dispersed throughout the town. In Ijebu cognatic descent groups are also land-holding units, and a person can obtain land through membership of several different groups. In Ondo, according to Lloyd, land is vested in the ruler on behalf of the community rather than in descent groups, and any native of the kingdom can farm anywhere within it. In both kingdoms the cognatic system has a strong agnatic bias, both in residence, and, in the case of Ijebu, in land tenure.

Bender (1970) took issue with this view of Ondo. He argued that it is the folk image of a people which ultimately determines whether a society is characterised as having a particular type of descent. The Ondo have, he argues, both a patrilineal ideology and agnatic descent groups. Pace Lloyd, an Ondo man is only a member of one descent group ñ that of his father. He also took issue with Lloyd's view of Ondo land tenure. Land is in fact vested in individual descent groups rather than in the ruler on behalf of the community. A man is normally only entitled to farm the land of his father's descent group. He may farm land allocated to his mother by her descent group and worked during her lifetime, and may even pass this land on to his children, but this, he argues, is also possible in other kingdoms like Oyo whose agnatic kinship system has never been questioned. Bender does admit, however, that the senior chiefs in Ondo gave him an account of Ondo land tenure similar to that of Lloyd, and he leaves open the question of how this difference between the chiefs' view and the Ondo people's view could have arisen.

This debate raises two major issues. First, if we are to adopt a typological approach to Yoruba kinship systems, on what criteria are we to distinguish between the types? Bender stresses folk ideology, while Lloyd stresses corporate functions and the transmission of jural statuses. This resolves itself to the problem of emic versus etic approaches, or the actor's frame of reference versus that of the observer. From the point of view of comparative analysis, it seems logical to take Lloyd's position, though in the case of the Yoruba this is problematic. As Lloyd himself makes clear (1962: 33), the corporate functions of Yoruba descent groups vary widely from area to area. Classifying Yoruba kingdoms on the basis of the presence or absence of 'agnatic descent groups' makes little sense if the specification of what an agnatic descent group actually does differs so widely. Even their role in the control of land is not at all uniform. Secondly, a typological classification of Yoruba kinship systems obscures the degree to which those systems are continuing to evolve. Elsewhere Lloyd clearly recognises these changes, and in his brief reply to Bender he sees this as one answer to the problem of Ondo land tenure. In a situation where farmland has taken on a new value for cocoa production and where pressure on it is increasing, descent groups may assume new functions with respect to land and a stronger corporate identity. In any case, corporate descent groups are contingent on other factors: they cannot be taken as the primary units of analysis in Yoruba kinship. Their development has to be explained where and when it has occurred.

Yoruba kinship: an alternative perspective

In the first part of this chapter I have attempted to show that some of the conventional theoretical approaches to the study of Yoruba social organisation have laid the emphasis on elements which are not universal, and that they have been unable to take account of the constantly evolving aspects of Yoruba society. In the rest of the chapter I attempt to develop an alternative approach, starting with four propositions. The first is that Yoruba kinship ideology has important bilateral elements in all areas of Yorubaland, and not just in Ijebu and Ondo. Secondly, despite these bilateral elements, Yoruba kinship in practice has a patrilineal emphasis in most areas, owing to two factors: the virilocal pattern of residence on marriage, and the close economic cooperation between father and son.

Thirdly, co-residence rather than genealogy may be crucial in understanding some aspects of social structure. This was implicit in the earlier focus of Johnson and Fadipe on the compound as the major unit of analysis, but a similar position is also taken by Bender in his paper on households in Ondo (1972). Households, he argues, exist in part by virtue of living together. They are best seen not as localised families, but as groups of coresidents, some of whom are related.

The final point is the dynamic, constantly evolving nature of Yoruba society. Differences in social organisation in different areas are the result of historical processes, and a response to political and economic factors. The actors themselves may put forward different versions of their kinship ideology to support their own interests, and which one becomes generally accepted will relate more to the present distribution of power than to questions of historical truth.

The rest of the chapter explores the relevance of these ideas to Yoruba kinship ideology, patterns of residence, inheritance, land tenure and marriage. In the final section they are related to settlement patterns and contemporary forms of urban development in Yorubaland.

Kinship ideology

The cognatic elements in Yoruba kinship organisation fit well with the Hawaiian kinship terminology.[2] This makes no distinction between agnates and cognates and largely ignores sex in both ego's and descending generations. The basic kin terms are limited in number: baba, 'father'; iya, 'mother', ,egbon, 'senior sibling'; aburo, 'junior sibling'; oko, 'husband'; aya (or iyawo), 'wife'; ana, 'affine'; and omoi, 'child'. These terms can be used in combination to specify relationships more precisely, and are regularly used in a classificatory sense to include distant kin, or even nonkin with whom an individual has quasi-kinship relations. English kin terms have also been borrowed: mama, dadi, broda ('senior brother') and anti (derived from 'aunt', but meaning 'senior sister') are all widely used.

The Yoruba terminology relating to lineage, lineage segments, households, families and ego-centred kin groups is similarly vague and imprecise. In the towns of northern Oyo the main groups with which individuals identified were compounds, ile, rather than descent groups or other kinship groups. The normal Oyo Yoruba term for descent group is idile, which literally means the 'stem' or 'root' of a house, though a person in Igbeti or Ogbomoso would be far more likely to talk about awon ara ile wa, 'the people living in our compound', or awon omo ile wa, 'the members of our compound (by birth)', than about his idile. Descent groups in other areas, including Osogbo, Ekiti and Ijqbu are called ebi (Schwab,1955; Lloyd, 1962). In Oyo as well, this term, which is derived from bi, 'to give birth', could be used to describe a descent group, though it has a much more general meaning and simply refers to a group of cognates. Yoruba literate in English often translate it as 'family' and its meaning is just about as vague. Idile has a more precise meaning, but is less commonly used. This is also the case in Ondo where an idile is a constituent part of a larger ebi (Bender, 1970: 76).
There are no standard terms to describe lineage segments. In Igbeti or Ogbomoso, people were more likely to be concerned with a particular house within a compound, rather than a descent-group segment. Ile can also refer to constituent houses or their residents, though kaa (aliter kara, ikara, ikaa) which literally means 'courtyard' is often used instead. In discussing inheritance and succession perhaps the most important term is omoiya, a group of children with a common mother and by extension, their descendants. On a man's death, the property he has acquired himself passes in equal shares to each group of omoiya, irrespective of the number in each group. Other terms used in the same context are origun or 'corner', used in both Oyo and Osogbo (Schwab, 1955; Sudarkasa, 1973); idi or 'stem', used in Ondo, Ijebu and many other kingdoms; and ojumu, used in Ijebu. In Osogbo, according to Schwab, there is a term isoko used to describe the group of agnatic descendants of a single man.

What gives Yoruba kinship its strong patrilineal emphasis is of course the pattern of residence after marriage. A woman moves to her husband's compound, and often remains there even after his death, marrying another member of his descent group or remaining with her children. However, as the great majority of women marry within the same town, they are able to maintain close ties with their own compounds, together with their rights of inheritance there. Some move back to their natal compound permanently, particularly if, on the death of the husband, they decide not to remarry. Their children may go with them with the result that, even in areas like northern Oyo a small proportion of men spend most of their lives with their mother's kin. They may only leave when they want to build a house, which they normally do on land belonging to their father's compound. Some men never acquire the money to build: they remain with their mother's kin and their children are increasingly regarded as belonging to the compound of their paternal grandmother.

Even in those areas where the cognatic elements in the kinship system are perhaps most marked, however, patrilocal residence is still the norm. Matrilocal residence may have quite different implications depending on the settlement pattern. In Ijebu, the small size of the settlements outside the capital means that they often consist of a very few descent groups. Thus a large proportion of the marriages must involve spouses from different settlements. Thus a man who wants access to land belonging to his mother's descent group may well have to change villages. Even so, in the example Lloyd cites (1966a: 494), 60 per cent of the adult men were still living in the settlement of their father's father.

Patrilocality in Ondo is even more marked. The most unusual feature here is that descent groups are not localised, but usually have members scattered throughout the town. The head of a descent group often lives in the same place as did its founder, but the compounds are small, and in each generation the younger sons move out to build elsewhere. Why this pattern should have developed here is not clear. It may have something to do with the evacuation of the town in the 19th century and its resettlement under British auspices: this would have inhibited the growth of the large warrior compounds which appeared in most other towns during this period. It may also be, in part, a function of a chiefship system in which titles were not vested in descent groups, and where promotion from one title to another was possible. Senior chiefs lived in title houses, into which they moved with their dependants on accession. When a chief died or was promoted, the residents had to move out to make way for the next chief and his household. Such a pattern would tend to inhibit the localisation of descent groups. Significantly, one of the few localised descent groups in the Ondo capital is that of the Jomu, the only senior chief whose office was traditionally vested in a descent group (Lloyd,1962: 109).

A principle of crucial importance in Yoruba social structure is that of seniority (cf. Bascom, 1942). Among members of a compound or in a group of cognatic relatives, seniority is defined in terms of birth order. A woman's seniority in her natal compound is defined by birth order, but her seniority in her husband's compound is defined by order of marriage. She is junior to all wives who married in before her, and to all children born in the compound before her arrival. Yoruba kinship terminology, which largely ignores sex, does differentiate on the basis of seniority. It is particularly important that juniors use the correct forms of address to senior relatives, and, in the case of wives, to co-wives and affines. Many older Yoruba do not know when they were born, but they do know precisely who is senior or junior to themselves. But it is not just a question of respect and deference. The junior members of the compound are expected to take on the dirtier and more onerous tasks.

The main authority within a compound or descent group thus lies with the elders, and the head is normally the oldest male member. In some areas he has the title of bale (baba ile): in others he is known as the olori ebi (head of the ebi). Traditionally the bale was the ultimate authority within the compound in matters of discipline, dispute settlement and the allocation of rooms, work and land. The elders in each constituent house had a similar authority over their own group of residents, and they would be able to settle minor disputes within it. The bale also acted as the link between the compound and the political authorities in the town, and was responsible for tax collection. If the oldest man was senile, his role was performed by the next most senior. The bale usually did not farm, but was supported by junior members. He presided over the meetings of compound or descent-group members which are still important in many areas of Yorubaland, particularly when their members control resources like farmland or building land and housing within the town. In some towns descent groups have virtually become land management agencies. A good example is the Awosika Family Association in Ondo which has an executive council and which has produced a printed constitution and history. It sells land to outsiders, manages a school, collects regular dues from its members, and helps them with the costs of education and funerals (Lloyd, 1962: 104; Bender, 1970: 76). 'Family meetings' may be attended by women, but more regularly by men, and the discussions are increasingly dominated by the younger more educated members rather than the elders. Most people attend the meetings of only one such group, apart from in Ijebu where it appears a large proportion of people regularly attend meetings of the descent groups of their other grandparents, in addition to that of their father's father (Lloyd, 1966a: 495).

The authority of the bale and other elders has been reduced with increased mobility and the diversification of the economy. With declining rates of polygyny the older men no longer monopolise the available women, and younger men are no longer as reliant on their elders for the money to organise their first marriage. Yet the responsibilities of age are still taken seriously. Elders are still given considerable respect and deference in public, and men with higher levels of education, well established in public life, will still prostrate before them. The elders play a crucial role in the organisation of rites of passage in which their kin have to be represented. Guyer (1972: 183ñ 9) shows how the major responsibility both of attending these events, and of collecting the necessary cash contributions expected, fall on the elders. If they are unable to get the full sum from their juniors they have to make it up from their own pockets.

A point which emerges from Guyer's discussion is the degree to which rites of passage involve groups of cognatic kin (ibid: 175ñ6). The main group mobilised for a ceremony are ebi, kin by birth, to the main celebrant, and for this purpose descent is traced through both males and females. Thus in the case of a funeral in Ibarapa, where she worked, the group relevant to a particular ceremony is made up of descendants of the grandparents of the deceased: the older the deceased, the larger the group of people likely to be involved. On these occasions, an elder acts as the representative for the entire group of his children and grandchildren, and in some contexts, for the members of his full sibling group and their children as well.

In the case of Yoruba migrants in northern Ghana, the pressure for elders to return home to their natal compounds in old age, even if they had spent most of their lives abroad, was extremely strong. It was normal for the older migrants to retire to Nigeria and to leave their property and businesses in Ghana to their junior relatives in order to take up the responsibilities of age at home. The Yoruba proverb 'when there are no elders, a town is ruined' (agba ko si, ilu baje) is one to which many Yoruba would still assent.

Inheritance and land tenure

It is in the transmission of rights to property and land that the cognatic aspects of Yoruba kinship become extremely significant. These patterns have been modified considerably in the last century with changes both in the type and extent of property to be transmitted, and in the categories of kin likely to share in it. As the question of land tenure will be discussed in detail in the next chapter, this section concentrates on other types of inheritance. The general rule is that property only passes between blood relatives: a woman can only inherit a share of her spouse's property if she married under the marriage ordinance. Otherwise husband and wife do not inherit >from each other and the property of a woman who dies without issue goes to her own kin, usually her full siblings.

A major question in Yoruba inheritance in the past was that of the relative rights of the siblings and children of the deceased (Lloyd, 1962: 282ñ300). In the 19th century, the estate of a well-to-do man might have included slaves, horses, money, an extensive wardrobe and pawns, in addition to rights over wives and children. Originally his main beneficiaries would have been his full siblings. In 1858, a decision of the Ibadan chiefs apparently increased the rights of the children against those of the siblings, though according to Fadipe the old pattern was still regarded as the ideal in Ijebu into the present century (1970: 140ñ6; cf. Awolowo, 1960: 33).

The present rule is that the property a man acquires himself passes to his children, while the property he inherits passes to his siblings. His status as a member of his own sibling group passes to the next most senior sibling or to the eldest son of the group of full brothers, while his status as head of his own nuclear family passes to his eldest son. When he dies, his personal property, such as houses he built himself on his own land, or cocoa farms he cultivated himself, becomes the 'family property' of his descendants, and the eldest son succeeds his father as administrator of it. On the death of the eldest son, the role is taken over by the next most senior son, and so on. If the group gets too large, and the property involved is substantial (for instance, a large number of houses) it may be partitioned between the groups of full siblings among his children or their descendants. After partition, other segments of the descent group cease to have any claim on it.

In the case of property divided between a man's children, the division is usually per stirpes, i.e. on the basis of groups of full siblings, rather than per capita (Lloyd, 1962: 297). Thus if a man has three wives, all of whom have produced children, the property is divided into three equal shares, irrespective of the number of children in each sibling group. Whether the mother was a legal wife or a casual lover is also immaterial, as long as the children contribute to the expenses of their father's funeral. The cognatic element in Yoruba inheritance arises from the rights of women to their parents' property. Women generally have inheritance rights similar to those of men where possible. With more and more wealth invested in assets like cash, property and cocoa farms from which women can benefit, these rights have been increasingly exercised. In addition, many women are wealthy in their own right, through trade. As the property of a wealthy man is likely to be divided up between the groups of his children by a number of wives, while women pass the bulk of their property on to their children alone, it is common for an individual to inherit more from his mother than from his father. In areas like Egba where land rights tend to be partitioned among individual farmers, the land rights of a man with no sons will pass to his daughters. Thus a farmer may often inherit land rights from his mother. The result is that both property and land transmission have taken on a strong cognatic appearance, even in some areas usually thought of as having an agnatic kinship system.

Residence, descent and marriage

Both cognatic kinship elements and, to a lesser extent, co-residence, are significant in Yoruba rules of exogamy. The basic rule seems to be that marriage is prohibited between partners who can trace a blood relationship, however distant. What this means in practice depends on the length of people's genealogical memories. According to Lloyd it means that in most areas you cannot marry anyone belonging to the descent groups of your four grandparents, while in Ijebu and Ondo it means that you cannot marry a partner with whom you share a common ancestor within five or six generations back (1966a: 488). These are approximations rather than absolute rulesñmany people simply cannot trace their genealogy back that far. In Ondo, according to Bender, a marriage will be permitted if no relationship is known, but it will have to be dissolved if one is discovered later (1970: 79). Fadipe (1970: 71) however suggests that even in the 1930s the rules were being less strictly applied than they had been before, and in Igbeti some informants suggested that a very distant relationship might be ignored unless the parents were against the marriage.

But in addition in Igbeti marriage between partners from unrelated descent groups co-resident within the same compound is forbidden. This apparently remains the case for some time after the constituent descent groups have separated to form separate compounds.

Logically, these rules of exogamy complement those of seniority. A woman who is married to a blood relative or to a man born in the same compound would occupy two incompatible positions of seniority: one defined by her birth, and a junior one defined by her marriage. A rule of descent and residential exogamy avoids this.

The normal age of marriage for men is between 25 and 30, and that for women between about 17 and 25. A man's father is still often responsible for arranging and financing his sons' marriages. A man whose father dies before he is married may well have to wait until he can raise the necessary money. The main variable in the case of women is education. Betrothal periods have gradually shortened. At the turn of the century, girls were betrothed in childhood (Fadipe,1970: 70): some were even promised to friends of their father before they were born (Bascom, 1969a: 59). A correlate of long betrothal periods was that a large proportion of the women were monopolised by older men, and the age of first marriage for men was correspondingly high.

There were variations in marriage customs between towns, and even between compounds, but a fairly general pattern emerges from the literature. First marriages used to be (and often still are) arranged by the parents. Lengthy but discreet enquiries were made, and partners from families with reputations for dishonesty, debt, hereditary diseases, witchcraft and the like were avoided. When they were satisfied, the boy's family would make an approach through an intermediary, and the If a oracle was consulted: the usual way of turning down a suitor was to claim that Ifa's response was unfavourable. If favourable, the ceremony called isihun or ijohun and the first of a series of marriage prestations established contractual obligations between the groups, and gave the groom exclusive sexual rights over the girl.[3] The marriage, however, was usually not consummated until her arrival in his compound several years later, and great importance was attached to virginity. During the betrothal period, further prestations were made, and the man and his friends were expected to provide farm labour for the girl's father. The final prestation, usually called idana, was made before the transfer of the girl to her husband's house. It included items like bitter kola, orogbo, symbolising long life, Malaguetta peppers, atare, signifying fertility, and honey, oyin, symbolising a happy family life, together with schnapps, beer, and a purse with money in it. In a Christian marriage it became customary to include a Bible and ring.

More recently, the majority of young people choose their own partners, though most take care to obtain their parents' consent. Virginity is less important, though some men see it as an ideal. Many couples start to have regular sex before the final transfer of the bride, who is frequently pregnant by the time of the marriage. The sexual rights of the husband during the betrothal period on the other hand are still important. A cause celebre in one of the towns where we worked concerned a man who had an affair with the fiancee of a member of the same church society. He was virtually ostracised by the others in his age-group in the town when it came to his own wedding.

The government publishes schedules of the maximum amount in each category of marriage prestation which is recoverable in the case of a divorce. In examples published in the mid-1970s, the total sum amounted to N100. In real income terms marriage prestations are probably worth less than they were during the early colonial period, and they may be considerably less than the husband spends either on the wedding celebration or on trading capital to set his wife up in business. The actual celebration may cost very little, but it can become a major occasion for conspicuous consumption, with hundreds of naira being spent on food, drink, the wife's trousseau and payments to musicians.

Affording to Fadipe (1970: 90) divorce in Yoruba society used to be very rare. Senior relatives would put pressure on the couple to reconcile their differences. Grounds for divorce included the laziness, indebtedness, dishonesty or insanity of the husband, but not his unfaithfulness or impotence. In the latter case, a substitute could be found by the descent group. In the colonial period divorce became more frequent, and petitions were granted almost automatically on repayment of the husband's marriage expenses.

The frequency of divorce in present-day Yoruba society and the degree of variation between particular areas and social classes, is not at all clear. In some areas it may be relatively frequent. Lloyd (1968a) estimated that Ijebu marriages are among the least stable in Africa, with perhaps 5 per cent of the marriages dissolved annually. In a study carried out in Ibadan by contrast, Olusanya (1970) found that only 10 per cent of his sample had had previous partners.

The ease of divorce varies with the legal status of the marriage. A small percentage of marriages, mainly of educated couples, are monogamous unions contracted under the marriage ordinance. These are unions which give the spouses rights roughly equivalent to those in the United Kingdom, and can only be dissolved in the High Court. The great majority of marriages take place under customary law which permits polygyny, and they can be dissolved in the local courts very easily. Few Yoruba men sue their wives for divorce, and when they do it is almost invariably on the grounds of adultery. A man tired of his wife simply stops supporting her. She leaves for a lover's or her parents' house and starts proceedings from there. New partners are easy to find, as it costs less to marry a divorcee than a first-time bride, and there is little stigma attached. In the study by the Okedijis in Ibadan (1966), the most common reasons given by the women for divorce were financial neglect (71 per cent), trouble with co-wives (32 per cent), trouble with in-laws (20 per cent) and lack of children (20 per cent). The infidelity of the husband is seldom given as a reason. Olusanya's study found that marriages contracted when the girl was younger than 20 were less stable, and that ordinance marriages were more stable. Polygynous marriages are less stable than monogamous marriages, and the frequency of divorce is inversely related to education. Both he and Lloyd found that divorce was not obviously related to infertility: many women seeking divorce were already pregnant by their new partners.

The main issue in divorce proceedings is the repayment of the expenses incurred by the husband, including marriage prestations and trading capital, but not clothes and food. In Lloyd's study, the amount actually repaid in the early 1960s varied from an average of £7 in Ijebu to between £30 and £50 in Ondo. Repayments vary according to the length of the marriage and the number of children it has produced. The longer the duration and the greater the number of children, the smaller the amount. Significantly an increase in the maximum amount repayable in Ijebu Ode and Ondo was followed by a decline in the number of cases coming before the courts (1968a: 78).

Traditionally, a Yoruba woman went through the marriage ceremony only once: there were no rituals to mark her remarriage after the death or divorce of her husband. The marriage did not end with the death of her husband. She would marry a junior member of his descent group and remain in the compound with her children. This had the advantage that she would retain her seniority there. A woman who did not want to marry anyone in her husband's compound could repay the marriage prestation, and either move back to her parents' house or marry anyone else she chose. In the northern Oyo towns, widow inheritance is still common in the more conservative compounds. For women past childbearing the question is less important. Lloyd has suggested that it is more common for women in Ijebu and Ondo to return to their parents' compounds than it is in other areas.

Kinship and settlement patterns

The approach to Yoruba kinship adopted in this chapter can be summarised as follows. First, the classification of kinship systems in different kingdoms as agnatic or cognatic should be rejected. There are so many variables involved, that the criteria for classification must be largely arbitrary, and such a typological approach cannot deal with systems in constant evolution. Second, the present wide diversity of kinship organisation results from the interaction of two or three basic principles of social organisation which the actors use to organise their lives. The first is that kinship is reckoned bilaterally. The second is that residence on marriage is virilocal, and that therefore, other things being equal, the outcome will be a group of agnatically related males living together, along with their wives and children. The third is that where groups of people who are unrelated happen to live together in the same social unit, a way has to be found of organising their joint social life, and this is achieved through the extension of the patterns of authority, seniority, kinship terminology and exogamic restrictions which are found in kinship units. The variations in the relative strengths of these principles can be related to a range of historical, environmental, economic and political factors, and also to the variations in Yoruba settlement patterns.

In the 19th century, conditions existed over a large part of Yorubaland which assisted the strengthening of the second and third of these principles. In Ibadan in particular the process of migration and resettlement led to the formation of extremely large and heterogeneous households under the leadership of the military commanders. Agnatic links tended to be strengthened as sons inherited their fathers' wealth and military roles. Over time, residential groups became organised as quasi-kinship groups, and then, as the forest was recolonised, they took over land-holding functions as well. Their common interest in large tracts of land meant that these groups retained a strong corporate identity. In general it seems likely that the degree of corporate identity is correlated with control of resources. In Ondo in the present century, the enhanced value of particular resources has led to a sharp increase in corporate activity of the descent groups that control them, and an increased emphasis on the principle of agnatic linkage in place of the cognatic principle which had previously governed control of land.

In Ijebu, on the other hand, the original decentralised settlement pattern survived, and this is the area where the agnatic principle is least strong. It seems a plausible hypothesis that, given the settlement pattern, cognatic rather than agnatic links are likely to remain strong. A dense network of cognatic relationships connects the Ijebu villages with each other and with the capital. The economic resources and opportunities available in the villages are limited, and if an individual wants to improve upon them, he must look to his cognatic kin elsewhere. The marked division of labour between the capital and the rural settlements only intensifies this process. In areas like Ibadan on the other hand, a process of rural colonisation with temporary farming settlements has taken place. Their residents remain members of social groups in the capital, and the need for them to turn to cognatic kin for economic opportunities does not arise.

Increasing scarcity of land in general appears to strengthen agnatic linkages. As we have seen, this appears to be the case in both Ibarapa and Ondo, but Lloyd (1970) suggests that it may be part of an even more general trend. Even where rights to land have generally been passed on between agnates, it has usually been easy for a sister's son of the landowning group to obtain land through 'begging'. According to Lloyd, in some places where land is scarce some people argue that their sisters' sons should get land from their fathers' people, and that their own land should be more strictly reserved for their own use and that of their male descendants.

There are other developmental processes which have acted to strengthen the cognatic element of kinship organisation. As women have accumulated wealth in their own right, kinship links traced through women have hecome more significant. The process has gone furthest in Egba and Ijebu. In both areas there are a number of very wealthy women involved in large-scale trade in Lagos and Ibadan. In Egba the partition of land and the development of a land market have gone further than in other areas, and the transmission of rights in land through women has become increasingly common.

Finally, it might be argued that geographical mobility and economic differentiation are also likely to increase the relative importance of cognatic kinship links. In Ghana a significant percentage of migrants were recruited by their matrilateral relatives. Fostering in Yoruba families is very common. Children are frequently sent to live with close relatives who need their help, or who can teach them special skills, particularly grandparents or parents' full siblings. With many Yoruba scattered throughout West Africa, fostering has taken on a new significance. An Ogbomosho boy born in Ghana may well be sent back to live with his grandparents in Nigeria while he attends primary school, then to return to Ghana for secondary education, before finally joining a mother's brother in Jos in order to find a job. Children sent home provide help for their grandparents round the house, while they become fluent in their own dialect of Yoruba. The strength of the sibling bond, and the wealth of many women, means that matrilateral links are commonly used.

Other forms of association in Yoruba towns

Cross-cutting the ties of residence and descent in the Yoruba town are other institutions, based on age, religion and occupation. The age sets, title associations and traditional religious cults have in most towns either gone into decline or disappeared. Their social functions have been taken over by the more recently-formed egbe which are found in nearly every town. The term 'egbe' used to refer to age sets etc. Now it is used to refer to any type of association, from formal bodies with elected officials to small informal groups of friends. In Igbeti and Ogbomosho the most important of these are based on religion, and a large proportion of the people in the town belongs to one or other of them. An egbe usually starts as a small informal group of friends from the same neighbourhood, of the same age, sex and religion. Some of these develop into more formal organisations with a name, a larger membership, and elected of ficials. They hold regular meetings, arrange dances and choose a particular type of cloth as their uniform (aso egbe) during religious festivals and rites of passage. A person's closest friends are often members of the same egbe.

Secondly, in the larger towns there are the occupational egbe. In Ogbomosho, for instance, there are over thirty of these, covering both market trade and the crafts, and ranging from tailors' and photographers' associations to cloth-sellers and the makers of local soap. Each has its own officials and regular meetings. Some of them attempt to regulate prices and quality, while the traders' egbe act as pressure-groups to get market amenities improved or taxes and licence fees reduced. Many of them run esusu rotating credit associations, and some of them make welfare payments to their members or help them with the expenses of namings or funerals. They are also important in settling disputes in the market. In Lagos there is an elaborate hierarchy of market officials headed by the Iyalode to deal with problems of this sort (Baker, 1974: 223ñ43). The Iyalode is the most senior of the women chiefs in many towns, and, as the majority of market traders are usually women, trade is often one of her main concerns.

Many towns have some form of parapo or town improvement union. Membership is open to all members of the descent groups in the town, whether they are living at home or elsewhere. The town unions developed in the 1920s and 1930s, both among the Yoruba and elsewhere in Nigeria. They were first organised by educated migrants in Lagos and the other large towns, and they have gradually spread. The membership of the outside branches consists of migrants from the same town living in the area: they have elected officials, written rules and regular payment of dues. In some cases, delegates from each branch return home for the annual general meeting. In northern Ghana, the Yoruba migrants received their mail through post office boxes rented by the parapo. At each meeting, messages and letters from the home town and from other branches were read out. The dues were spent on running expenses and on maintaining good relations with the local officials. There were also special collections towards development projects at home ñ a new church or mosque, a town hall, a dispensary or a new market site. In the 1930s the unions, representing the new commercial and educated elite, started to play an important role in local politics and some of the officials later became elected councillors or party politicians. In the 1950s some of the functions of lobbying the administration were taken over by the political parties, but with the advent of military rule the town unions revived in importance. Delegations of parapo officials calling on high-ranking sons of the town in the civil service or army to make their views known are common, whether the aim is to get a new hospital, secondary school or piped water supply at home, or to ask for intervention in a land or chiefship dispute. Thus the activities of these unions are directly related to the unequal levels of development in different areas, and the allocation of amenities by the government.

Conclusion: social change, kinship and the home town

It is these inequalities which raise the most interesting questions about Yoruba towns at the moment, even though most writers have concentrated on historical and definitional problems. Towns that are growth poles have high rates of immigration and unemployment: those that are not suffer from loss of manpower and economic stagnation. Thus the development patterns and the problems of Ogbomoso, Saki and Iwo are very different from those of Ilorin and Osogbo, despite their original similarities of size and social structure. These differences are reflected not only in growth rates and migration statistics, but also in physical layout, land values and occupational distribution. Among the Yoruba, the rural-urban dichotomy takes an unusual form, with some relatively large settlements suffering from the typically 'rural' problem of economic stagnation, as they act as a labour reservoir for the major economic growth centres.

The extent to which migration to Lagos, for instance, is draining off a considerable proportion of rural manpower from some areas is clear from Green's data (1974). In the years between 1952 and 1963 Lagos grew at a rate of 8.6 per cent per annum. This included the movement of 510,000 people from other Yoruba areas to Lagos, mainly from the surrounding parts of Lagos State and the southern divisions of Ogun State. A second stream of migration has been to the eastern areas of the cocoa belt ñ to Ondo, Ekiti and Owo. The towns of the older cocoa areas and the savanna have lower population growth rates. By 1963, a list of the ten largest towns in Nigeria included only four in the Yoruba area: Lagos, Ibadan, Abeokuta and Ife (1974: 282). The pattern of urbanisation among the Yoruba is now much more similar to that in the rest of Nigeria: a single large growth pole attracting manpower away from a hinterland with slow-growing or static populations and declining agricultural productivity.

Kinship links play an important part in channelling these migration streams, as people move to join their relatives in other towns in order to find jobs. In the other direction, there is a constant flow of money, goods services and people between the established migrants and their relatives at home. For most people, the compound at home still represents the final security. If all else fails, a Yoruba migrant can return home and farm. There have been two major events in recent years which resulted in large numbers of migrants having to return to their home towns: the disturbances which preceded the Nigerian civil war in 1966ñ7 (Sofola, 1971) and the expulsion of the West African aliens from Ghana in 1969 (Peil,1971; Hundsalz, 1972).[4]

This realisation that security lies at home also helps explain why Yoruba migrants retain their cultural identity so strongly in the areas in which they settle. In northern Ghana, the Yoruba usually lived with their relatives. They imported their wives from the home town, and formed virtually endogamous groups. The elders all planned to return home eventually to assume positions of responsibility in their own compounds. The wealthy migrants had usually built a house at home as one of their first priorities, and were able to remain in touch with events there through relatives and the town unions.

Finally, this loyalty to the home town is relevant to Yoruba politics, as towns within a district and districts within a state compete for resources, using their links with members of the civil service and the military. But all these processes are ultimately linked to patterns of economic change, and it is with these that the next chapter is concerned.