History and Social Structure: A Study of the Sefwi Residential System (Ghana)

Article excerpt

The article reviews studies concerning the Akan (Ghana) residential system with particular reference to Fortes's Time and Social Structure. Fortes's work is criticized for its lack of historical perspective, for assuming the structural identity of kin groups, and for ignoring the political and economic importance of the father-child relationship. The author presents a survey of the Sefwi residential system in which domestic units are differentiated according to their historical role. The economic and political importance of the father-child relationship is acknowledged in the transmission of and rights and in the dynamics of household foundation. (Matrilineal kinship, residential system, history, Akan, Ghana)

The above title, a paraphrase of Fortes's famous Time and Social Structure: An Ashanti Case Study, was chosen because it expresses the main purposes of this article. The aim is twofold: to review the literature concerning the residential pattern of the Akan, centered on Fortes's (1949) work and assumptions, and to propose a different framework through the introduction of a historical perspective.

An analysis of Fortes's kinship studies of the Akan, with particular reference to their residential system, enables one to point out certain theoretical shortfalls common to British social anthropology in the 1940s and early 1950s. This is important for several reasons. The accepted theoretical framework in Time and Social Structure is particularly clear and its consequences are symptomatic. Moreover, this article may also be relevant to students of the Akan area since the theoretical assumptions and results of Fortes's work (particularly Time and Social Structure) have often been accepted uncritically in many of the studies which followed. Finally, the principal aim here is to propose a different theoretical framework and offer insights which will shed new light on the residential pattern of the Akan.

The first two parts of this essay analyze Fortes's theoretical framework and his publications on the Asante. In the third section, Fortes's theoretical framework is criticized and a different methodological and theoretical procedure is proposed. Finally, the revised approach will be applied to the study of the residential pattern of the Sefwi.

FORTES'S ASSUMPTIONS

Fortes's theoretical assumptions and methodology have been clearly and repeatedly specified in his work and have been applied in his fieldwork among the Asante. It is therefore essential to examine briefly some of the main features of his kinship theory and methodology to understand his description of the Asante residential system. Fortes's fundamental assumption is that there are a few simple norms on which society is forged. These rules create a structure, a "social structure." The rules derive from the one law which gives society its structure: the principle of descent. Rights regulating access to land, the criterion of political citizenship, the system of inheritance, and everyday social relations all derive from the descent principle (Fortes 1953, 1958, 1959).

The residential pattern is not a direct product of the descent principle. Fortes admits the importance of affinity and time in determining the "form" and "structure" of the residential system.

[T]ies of kinship, marriage, and affinity regulate the structure of domestic and family groups, which have no permanent existence in time. Each domestic group comes into being, grows and expands, and finally dissolves. But the institutions it embodies, and the mode of organization it exhibits, are essential features of the social structure. Domestic organization has two aspects. Its form derives from a paradigm or cultural "norm" sanctioned by law, religion and moral values. Its structure is governed by internal changes as well as by changing relations, from year to year, with society at large. (Fortes 1949:7, cf. 1950:261, 1953:91)

The descent principle also determines the unit of study of the social structure: the lineage. Since society is forged by a sole principle (that of descent) and the units of study are a consequence of this principle, the units of study (lineages and segments of lineages) share identical characteristics. "The general rule is that every segment is, in form, a replica of every other segment and of the whole lineage" (Fortes 1953:85; cf. McCaskie 1995:169). Alongside the line of filiation which the group acknowledges as the descent line, the social recognition of another line, the "complementary line of filiation," is admitted but, according to Fortes, this is of secondary importance as it is not tied to the politico-jural domain.

It appears that there is a tendency for interests, rights and loyalties to be divided on broadly complementary lines, into those that have the sanction of law or other public institutions for the enforcement of good conduct, and those that rely on religion, morality, conscience and sentiment for due observance. Where corporate groups exist the former seem to be generally tied to the descent group, the latter to the complementary line of filiation. (Fortes 1953:89, cf. 1963:60-61)

The methodology of Fortes's works is characterized by the lack of a diachronic approach. Historical sources, both oral and written, are considered unreliable and therefore ignored (Fortes 1949:1-2). Fortes identifies three functions of time. Time as continuity "is significant as an index of forces and conditions that remain more or less constant over a stretch of time" (Fortes 1949:2). Time in this first meaning is viewed as static. When time is taken into consideration as a dynamic factor, as is the case in Time and Social Structure, the diachronic analysis is generational or, as Fortes calls it, genetic. It is concerned with the process of change from one point of stability to another. The time span considered is limited to the substitution of one generation with another. It is concerned with "replacement" cycles through a repetition of "stages." "Residence patterns are the crystallization, at a given time, of the development process" (Fortes 1958:1-3, cf. 1974a:8). Time is considered here in its cyclical significance. Nondynamic and cyclical time are introduced in Fortes's theoretical framework to cross out history, the third type of time, which is included by Fortes in the category of "duration time" and is dealt with as "an extrinsic factor having no critical influence on the structure of social events or organization" (1949:1).

ASSUMPTIONS AND RESIDENCE

Fortes's theoretical framework and methodology were applied to the Akan area. Akan are matrilineal and to Fortes this was the key to read their society: political citizenship was a consequence of membership to matrilineal clans and lineages; inheritance was matrilineal; land access was obtained through membership to a uterine group. The complementary line of filiation (in the Akan case the relation between father and son) was excluded from the politico-jural domain and confined to moral and spiritual relationships (Fortes 1963). The residential system was not included in the politico-jural domain. The pattern of residence was partly influenced by the bonds resulting from the complementary line of filiation, but it could not be completely separated from the general norms regulating the social structure (Fortes 1949).

A prevalent virilocal postmarital arrangement or a patrilocal pattern of residence would have contradicted the forging principle of the social structure. It would have been difficult to present as coherent a system which combined a social structure forged by a matrilineal principle and a virilocal-patrilocal residence. Such an arrangement would have led to the "matrilineal puzzle," the problem of combining exogamy and local descent continuity. On the other hand, Fortes could not maintain that the residential pattern was uxorilocal since all evidence contradicted this. He therefore proposed a rather unusual postmarital residential system termed duolocality: man and wife would live in their matrilineal households, children being brought up by their mothers (Fortes 1949:10-11). His data could not fully support this assumption, so a generational cycle was elaborated in Time and Social Structure. He presents a complex system of movements of people who form different kinds of residential units that could, eventually, be considered variations of the prevalent matrilineal household. Fortes's description of the residential system of the Asante is characterized by a combination of two factors: the conflicting claims of the agnatic-affinal bonds and matrilineality; and the time factor.

The first factor is explained as follows: "It can be suggested that the type of residential unit found in a particular case is a result of the balance struck between the obligations of marriage and parenthood on the one hand and those due to matrilineal kin on the other" (Fortes 1949:17, cf. 22-24). This conflict produces three possible forms of domestic units:

A. Households grouped around a husband and wife. In the simplest case this corresponds to the elementary family consisting of a man, his wife, and their children; but other kinsfolk may be included in the group.

B. Households grouped around an effective minimal lineage or part of it, such as a woman, and her sister or daughters, or a man and his sister or sister's son. C. Households made up of combinations of the previous types, e.g. a household consisting of a man and his wife and children as well as his sister's children. (Fortes 1949:16; cf. Fortes, Steel, and Ady 1947:168; Fortes 1950:262)

The existence of miscellaneous dwelling groups is recognized (Fortes 1949:16). Fortes believes that "one's mother's home--that is, by definition, the place where one's matrilineage is domiciled--is one's true home," but one may ask to have a room in the paternal domestic unit (Fortes 1949:17; cf. Fortes, Steel, and Ady 1947:166-67).

The next step of Fortes's analysis is to state that differences in the residential patterns of households are linked to the sex of the head of the domestic unit. Female-headed households will be likely to have a B-type structure, while male-headed households will have a minor share of matrilineal relatives (Fortes 1949:19-22).(2) In this section Fortes presents his data on household composition (Table 1).

Table 1 (Fortes's Table 4): Classification of Members of
Households with Male Heads (Excluding Heads)(3)

A. MALE MEMBERS
                                        Asokore

                                   Number       Per Cent
I. Lineage kin
Full brothers                        6            3.0
Full sisters' sons                  38           19.0
Classificatory sisters' son          3            1.5
Sisters' daughters' sons             7            3.5
Other matrilineal kin                9            4.5

Total I                             63           31.5

II. Nonlineage kin
Own sons                            94           47
Daughters' sons                     24           12
Other patrilineal kin               16            8
Miscellaneous kin                    4            2

Total II                           138           69

Totals I and II                    201          100

B. FEMALE MEMBERS

I. Lineage kin
Full sisters                        14       9(c) 6(d)
Full sisters' daughters             26       16   11
Classificatory sisters'             ..          ..
   daughters
Classificatory daughters'
daughters                            1       1   0.4

Other matrilineal kin                8       5   3(f)

Total I                             49      31    20
II. Nonlineage kin
Daughters                           83      52    36
Daughters' daughters                23      15    10
Other patrilineal kin                3       2     1

Total II                           109      69    47

III. Spouses and affines
Wives                               71            31
Wives' kin                           5             2

Total III                           76            33
Totals I, II, and III              234     100   100

A. MALE MEMBERS
                                         Agogo

                                  Number       Per Cent
I. Lineage kin
Full brothers                       15             7
Full sisters' sons                  56            27
Classificatory sisters' son         ..         .. (a)
Sisters' daughters' sons            30            14
Other matrilineal kin               16           8(6)

Total I                            117            56

II. Nonlineage kin
Own sons                            61            29
Daughters' sons                     10             5
Other patrilineal kin               13             6
Miscellaneous kin                   10             5

Total II                            94            45

Totals I and II                    211           100

B. FEMALE MEMBERS

I. Lineage kin
Full sisters                        24     9(c)    8(d)
Full sisters' daughters             71     26     23
Classificatory sisters'             18      7      6
   daughters
Classificatory daughters'
daughters                           15      6      5
                                    {7     {3    {2(e)
Other matrilineal kin              {38    {14   {12(f)

Total I                            173     65     56

II. Nonlineage kin
Daughters                           64     24     20
Daughters' daughters                17      6      5
Other patrilineal kin               15      6      5

Total II                            96     36     30

III. Spouses and affines
Wives                               46            15
Wives' kin                          ..            ..

Total III                           46            15
Totals I, II, and III              315     100   100

(a) Mainly mothers' sisters' daughters' son

(b) Includes a number of mothers' brothers

(c) Percentages exclusive of wives and their kin

(d) Percentages inclusive of wives and their kin

(e) Mostly more distant than mothers' mothers' descendants

(f) About half are mothers and mothers' sisters

The final part of Fortes's essay deals with time, the second determinant factor, with reference to the movement of wives and children from the woman's matrilineal household to the man's. Fortes's field survey shows that about half of the wives who are supposed to reside in their husbands' compounds as a consequence of a postmarital virilocal norm in reality live in their own matrilineal households. The duolocality of Asante marriage arrangements is thus numerically sanctioned. A development cycle and "genetic time" are introduced by showing that wives are more likely to live with their husbands in their thirties (Fortes 1949:24-25, 1950:262, 1954:270). The residential dynamics of sons and daughters are to be considered in relation to the sex of the household head. Households headed by women expel sons after marriage, while daughters of the household head (or of her sister) remain in the compound regardless of their "social maturation." Households headed by males, if the wife of the household head is present, allow some of the children of the household head to remain in the household even after reaching adulthood (Fortes 1949:24-31).

Time and Social Structure lacks the straightforwardness of most essays of British social anthropology of the period. A clear structural law governing household composition seems to be missing. Fortes evidently could not determine a clear link between residential norms and principle of descent. He was therefore forced to introduce several variables to attempt an explanation of his data. Time and Social Structure is concerned specifically with the analysis of the residential pattern of the Asante. Fortes's following works on the Asante reproduce, often in a simplified form, the concepts exposed above. The simplification is normally achieved by crossing out the time factor.

The household may be under a male head or a female head. If it is under a female head, it is normally a segment of a matrilineage consisting of the head and her children, her sister and her children, and perhaps her own and her sisters' uterine grandchildren. The household of a male head, on the other hand, may be either a parental family, consisting of a man and his wife or wives and their children, or it may include the head's sister and her children as well as his wife and his own children, with, sometimes, the children of his children or of his nieces. (Fortes 1950:261)

The subject is dealt with briefly in Fortes's works of 1954 (1954:270) and 1971 (1971:4), in which the same concepts are expressed. In 1970 the residential pattern is reduced, by crossing out the matri-patri conflict as well as the time factor, to the prevalent matrilocal-duolocal model:

The most common pattern in the forties was the dwelling group made up of the uterine descendants of both sexes of a common grandmother or great-grandmother, as frequently under a woman head as under a male head, in effect therefore a domestic segment of the maximal lineage. (Fortes 1970:163, cf. 1970:163 n. 22, 203 n. 24)

REVISED ASSUMPTIONS

Fortes's model was embraced by social scientists and its main features are restated in many publications concerning Akan kinship. Gough (1961) and Basehart (1961) reconstruct the Asante residential system based on Fortes's studies (Basehart 1961:287-89; Gough 1961). Bleek (1972, cf. 1975:30-38) and Woodford-Berger (1981) criticize some aspects of Fortes's description of the residential pattern, but both believe that a duolocal-matrilocal arrangement prevails. Christensen (1954), Hardiman (1974), Vercruijsse (1972, 1974), Oppong (1974), and Kyei (1992) interpret the Akan residential system on Fortes's assumption of a conflict between the pulls of the matrilineage and of the bonds produced by marriage and patrifiliation. Abu (1983) and Tetteh (1967) support Fortes's thesis of the importance of the separateness of spouses with data from their surveys carried out in the Ashanti and Brong Ahafo regions, which show that approximately half of the married women do not live in the same household as their husbands.

While Fortes's model has influenced a great deal of the research on Akan kinship, some scholars upheld different positions. The ethnographies which preceded Fortes's research described a virilocal-patrilocal model of residence. Rattray (1929:2-3, 7, 63, 1927:327) reports the prevalence of a bilateral parental pattern of household composition. Rattray (1929:22), Amoo (1946:231), and Kobben (1954; cf. Tellier 1902: 148) describe the residential arrangements respectively of the Ashanti, of the whole Akan area, and of the Agni as patrilocal and virilocal. Hill (1975:120) restricts the validity of Fortes's analysis to old cocoa plantations and states that a conjugal postmarital residential pattern is most common. Italian ethnologists working in the Nzema area have expressed a similar view (Signorini 1977:295-97; Palumbo 1991:171-75, 1992:252; Pavanello 1995:39). Akwabi-Ameyaw (1982) criticizes Fortes's generalizations but does not propose an alternative model.

Fortes's insufficient attention to the process of social change has also been challenged. While Fortes admits that some differences in the residential dynamics of the two villages chosen as the object of study are due to social change (namely, more marked patrilineal tendencies are present in the village with stronger urban contacts), he fails to include this variable in his theoretical discussion. Fortes (1949:22-23, 27, 30, 32) considers the variations between the two villages as a "local bias" which gives shape to different forms of the same structure forged by the dichotomy between matrilineality and marriage. On the other hand, the shift from a matrilocal-duolocal residence toward a patrilocal-virilocal one, which occurred as a consequence of the general process of acculturation and modernization, is the central issue addressed by Lystad (1959), Zajaczkowski (1960), and Vercruijsse (1972). Fortes (1971, 1974b), Bleek (1975:30-38), and Palumbo (1991), on the contrary, criticize the assumption that the process of modernization necessarily damages matrilineality.

Even though scholars of Akan kinship have expressed very different opinions on the residential system of the Akan, they share a common methodological assumption in the belief that one rule governs the residential system of all households. The contents of the rule vary according to the authors, but two tendencies may be identified. Some works follow Fortes's analysis and describe a dominant matrilocalduolocal pattern of residence, while others support the prevalence of a patrilocalvirilocal system. The present study shares Fortes's attempt to identify principles of social organization which forge the Sefwi, and more generally Akan residential system. I believe, however, that such principles are neither few nor simple. Akan kinship has been oversimplified by the consideration of the opposition between descent and marriage as the prevalent organizing principle. The aim here is therefore to contribute to a process of construction of a framework which enables approaching the complexity of Akan residence. This will be done by introducing a new variable-namely, the consideration of the rank of households--and by adopting a different methodological use of time. I argue that an analysis of the social and political status of each household is necessary to understand the functioning of the residential system. It is essential to reconsider Fortes's assumptions, and in particular his conception of time, to understand the differentiated rank of households.

It is well known to all students of the Akan area (and was noticed by Fortes himself) that not all those who live in a village have the same social and political status. This can be exemplified by examining the process of household foundation. The common oral tradition of the foundation of villages begins with a journey of a man and his sister. The couple arrives either in an uninhabited area, which is then colonized, or may reach a territory occupied by another group. In this latter case, the immigrants ask the political authority permission to settle and farm the land. The man and his sister, in the oral tradition, are presented respectively as the first chief, ohene, and the first queen mother, ohemma,(4) of the village and will normally be recognized as the founding ancestors of a matrilineal segment.(5) This uterine group is recognized as the royals, the adehyee (sing. odehyee), of the newly founded village. The authority over the inhabitants of the village, vested in the offices of ohene and ohemma, is transmitted within this matrilineage. The adehyee, through the chief, hold rights over the territory surrounding the village. The distribution of land to members of the adehyee's matrilineage and to other residents in the village rests upon the consent of the chief.(6)

A historical perspective, which goes beyond the simple life cycle of a single family unit, shows that the groups living in a village are perceived as having different ranks. The inhabitants of the village include the adehyee, the recognized matrilineal descendants of the founders, and others who, according to tradition, have settled in the village after its founding. I argue that the residential norms of groups that differ in social and political status are not uniform. The assumption that the units of study, the lineages and the dwelling units, are identical is misleading. To put it in Fortes's terms, the "duration time," the historical dynamics of a group, determines its behavior in the "genetic or growth process."

Another assumption which has been extensively challenged is the exclusion of the complementary line of filiation from what Fortes termed the "politico-jural domain." Sons have been shown to have held key political positions. Fathers throughout the Akan world had good reasons to invest power in their sons and, since the nineteenth century, used this kinship option for political purposes. Historical studies by Arhin (1983a:474, 1983b:9), McCaskie (1980:199-200), Perrot (1982a:192-215, 1982b:459-60, 1985), Terray (1976:312, 1987'36-41), and Wilks (1967) declare that sons of chiefs were appointed to key political offices throughout the Akan area. In the kingdom of Sefwi Wiawso, sons of kings occupy crucial positions within the political structure. The offices of Kontihene, Adontenhene, Wirempihene, Benkumhene, and Nifahene, together with other minor stools, were held by sons of deceased rulers. The relationship between the king, the omanhene, and these stool holders is still referred to as one between father and son (Boni 1997). While most evidence concerning the distribution of political power to sons refers to the assignment of offices within kingdoms, we shall see that similar strategies were adopted within villages. A critical examination of Fortes's work has provided new assumptions for the analysis of Sefwi residence in the use of a historical perspective, in a differentiated approach to the residential arrangements of single households, and in recognizing the political significance of the "complementary line of filiation."

SEFWI RESIDENTIAL PATTERNS

An analysis of the residential dynamics of two Sefwi villages, offered below, provides a different explanation of the Akan residential pattern. Sefwi lies west of Asante and is part of a wider Akan area which shares similar political, linguistic, and social features. The Sefwi area was populated by immigrant groups, mostly Akan, which merged with the autochthonous population throughout the last three centuries (Boni 1997). The two villages chosen for this study are just a few miles from the boundary of the "Ashanti Region." In the oman, the kingdom of Sefwi Wiawso, the residential pattern of different settlements varies considerably. The political history of the various settlements contributes to determine the norms of access to farm land and the pattern of land access influences the residential pattern. Some adekurofoo (sing. odekuro), chiefs at the bottom of the political hierarchy of the oman, had and have little power to regulate access to the territory surrounding their villages. The land around villages ruled by adekurofoo is administrated by a safohene (pl. nsafohene), an important chief, or by the omanhene. Large areas controlled by these important political authorities have been sold to strangers, mostly since the 1940s (Hill and McGlade 1957; Arhin 1986). Settlements in these areas are composed of unrelated nuclear households of immigrant farmers.(7) On the other hand, settlements founded in the eighteenth, nineteenth, and early twentieth centuries were incorporated into the Sefwi political structure. The matrilineage of the founder of a settlement, through its chief, holds the rights over the surrounding land. Some of these settlements have been scarcely influenced by recent immigration and the inhabitants are part of a network of relations of filiation. The validity of the data I present and the generalizations I attempt are limited to settlements of the second type. I concentrate on the residential dynamics of the descendants of two such settlement founders.

Anglo is a settlement founded at the beginning of the twentieth century by a group of immigrants which arrived from the Anyi area of the Ivory Coast. The group was asked by the omanhene to settle toward the boundary with the oman of Sefwi Anwiaso. The Anglohene, the chief of Anglo, was granted a certain degree of independence in the administration of the land surrounding his village, even though Anglo is an odekuro under the neighboring chief of Nsuonsua.(8) In 1993, when the survey of this village was completed, all but two of the households were inhabited by descendants of the village founders. The village numbered 296 inhabitants. Immigrant farmers were normally settling in smaller settlements surrounding Anglo (Figure 1). Suhienso lies not far from Anglo on the Asante border. The village was built approximately in the same period as Anglo by Asante immigrants and has been a recognized safohene since the 1930s. Suhienso also administers the land surrounding the village independently through its chief. The inhabitants numbered 278 in 1993. Out of the 22 households, eight were founded by foreigners (Figure 2).

[Figures 1-2 ILLUSTRATION OMITTED]

If the analysis of the residential pattern is limited to a single household, the historical interrelation between residential units is ignored. It is therefore necessary to examine the position of the founder of the household in the genealogical chart of the village to understand the residential pattern of single households. The genealogy of the residents in the villages of Anglo and Suhienso illustrates this clearly (Figures 3 and 4).(9)

It is necessary to distinguish the various residential units according to their historical origin and the genealogical role of their founder. The adehyee of the village are the matrilineal descendants of the founders of the village. They are the members of the royal matrilineage, the one in which the title of village chief is vested.(10)

[Figures 3 and 4 ILLUSTRATION OMITTED]

The founders of other households in Suhienso and Anglo are children (or grandchildren) of a male odehyee, normally of a chief or of the head of the matrilineage, an abusua panin (pl. mmusua panin). The son or daughter of a chief is referred to as an oheneba (pl. ahenemma). This term is a compound of the words ohene (chief) and ba (child). In the following pages I will extend the meaning of oheneba to all sons of male adehyee. Most founders of ahenemma's households are sons of chiefs or mmusua panin. We can therefore distinguish residential units of three types: those founded by an odehyee; those created by an oheneba; and those built by strangers. The three types of households follow different residential patterns.

The households founded by immigrant farmers are shaped around the nuclear family of the immigrant who acquired land in Sefwi. The household normally includes all other relatives, regardless of their parental collocation, who followed the first migrant. Adehyee's residential units normally consist of a matrilineal segment. The matrilineage tends to incorporate in its residential units dispersed members by recalling them, while sons of male adehyee are expelled from the unit when they reach adulthood. The matrilineality of adehyee's households is thus preserved over generations. The headship of adehyee's households is vested in the matrilineage. The return of matrilineal members, dispersed through virilocal marriage, to adehyee's households is achieved through the attribution of benefits and land rights to the returning members. The adehyee hold large tracts of land which can be granted as family land to some of the young, landless matrilineal members (Benneh 1970). Dispersed members of the lineage often inherit property and cocoa farms in the village where they hold the position of adehyee. Others receive gifts in land by matrilineal relatives. The matrilineage may call one of its dispersed members to reside in the household as chief, ohemma, or abusua panin. The adehyee normally form a large group and can therefore ensure assistance to the incoming members. Dispersed female members often return to their matrilineal household with their children when their marriage is interrupted.(11) Moreover, the matrilineage will encourage its women to marry in the area surrounding the village in which the matrilineage holds the stool, thus enabling them to visit their matrilineal household periodically and to return to the household in case of trouble with their husbands.(12) Finally, adehyee's matrilineal land may be used to attract husbands and evade the prevalent norm which establishes a virilocal postmarital residential model. The matrilineage will agree to assign, through the wife, even large parcels of land to foreigners who reside in the village after the marriage with adehyee women.(13) The land will pass from the immigrant to his children and thus return to the matrilineage. Even though some of the members of the matrilineage are made to return to the village where the matrilineage holds the stool, there are not sufficient resources to recall all members of the matrilineage. Some adehyee of the village will therefore reside in different (normally surrounding) villages. For example, an odehyee of Anglo married the chief of Sayerano and the children of the couple created some residential units in Sayerano. The adehyee of Anglo in Sayerano are ahenemma, while the ahenemma houses number 4 and 8 in Anglo were founded by adehyee of the village of Asawinso (Figures 1 and 3). Individuals therefore hold the different roles of adehyee and ahenemma in different villages.

The strategy aimed at the conservation of a pure matrilineal adehyee's household necessarily goes along with the expulsion of children of male adehyee. As postmarital residence is normally virilocal, sons and daughters of chiefs and of mmusua panin are brought up in the chiefs' household, the ahenefie. When they reach maturity, pressure is put upon them, as on all children of adehyee members, to leave the adehyee's compound. Children of adehyee have two choices: either they turn to their matrilineal kin in the village founded by their matrilineage; or they remain in their father's village. Their choice will depend upon the opportunities offered to them in both villages. Some sons and daughters of male adehyee, normally of chiefs and mmusua panin, find it advantageous to reside in their fathers' villages and will therefore build a new household, an oheneba's household, there. Ahenemma are therefore assigned a spot in the village to build their house. The newly built house is not defined socially for its matrilineal membership but is classified as the house founded by an oheneba. Even though the oheneba maintains a relationship with his/her matrilineage, often the matrilineal link slowly loses its strength due to the decreasing social contact and to the strengthening of the relationship between the oheneba and his/her father.

Residential units founded by ahenemma are often not considered matrilineal family houses, but buildings individually owned by the founder. Male ahenemma therefore tend to strengthen their relationship with wives and descendants. They, together with the father, represent the oheneba's social support in the village. In ahenemma's households, sons and daughters of the household founder will normally be allowed to live with the father and maintain their rooms after his death. Ahenemma may have nephews living with them, a matrilineal successor may come to inherit part of the land and a room. The successor will not, however, claim the whole house, and the major share of the deceased's land and wealth will be divided between his wives and children. At the death of the founder of an oheneba's household either a son or a matrilineal nephew may succeed as household head. Households founded by ahenemma will therefore include a major share of patrilineal descendants of the household founder. Their parental composition could be described as bilateral.

The above discussion proves useful for the analysis of the quantitative data on household composition. Fortes ignores the criterion of differentiation of households. He tries to elaborate a sole residential pattern applicable to all households regardless of their social origin and political status. Fortes's results show that matrilineal and patrilineal tendencies go together, but the data fail to clarify how the two tendencies are combined. I present below my data following Fortes's scheme, while adding the new variable of differentiation of households according to the status of the founder. The data from the Sefwi villages of Anglo and Suhienso, through the consideration of the historical-parental status of households, offer a clearer numerical picture than the one presented by Fortes (Table 1). Patrilineal and matrilineal tendencies tend to coexist in all households but their influence varies according to the household's origin. Adehyee's residential units are characterized by a prevalent matrilineal relationship between members, while ahenemma's residential units show more consistent patrilineal tendencies (Table 2).

Table 2: Classification of Members of Households in Relation to Household Head and Status of Household (Excluding Heads)(14)

                                       A. MALE MEMBERS

                                           Adehyee's
                                           compounds

                                    Number       Per Cent
I. Lineage kin
  Full brothers                          ..
  Full sisters' sons                      5
  Classificatory sisters' sons            3
  Sisters' daughters' sons                6
  Sons(*)                                 8
  Other matrilineal kin                   8
Total I                                  30            24
II. Nonlineage kin
  Own sons                                4
  Daughters' sons                         2
  Sons' sons                              7
  Brothers' sons                         ..
  Other kin                               4
Total II                                 17            13
III. Spouses and affines
  Husbands                                7            5

                                         B. FEMALE MEMBERS

                                       Number       Per Cent
IV. Lineage kin
  Full sisters                            5
  Full sisters' daughters                 8
  Daughters(*)                           10
  Sisters' daughters'
    daughters                            10
  Other matrilineal kin                  16
Total IV                                 49            38
V. Nonlineage kin
  Daughters                               2
  Daughters' daughters                   ..
  Sons' daughters                         4
  Brothers' daughters                    ..
  Other nonlineage kin                    5
Total V                                  11            9
VI. Spouses and affines
  Wives                                  13
  Wives' kin                              1
Total VI                                 14            11
Others                                   ..
Male and female totals                   128          100

                                         A. MALE MEMBERS

                                            Ahenemma's
                                            compounds

                                       Number       Per Cent
I. Lineage kin
  Full brothers                          13
  Full sisters' sons                     14
  Classificatory sisters' sons            2
  Sisters' daughters' sons                6
  Sons(*)                                10
  Other matrilineal kin                  10
Total I                                  55            18
II. Nonlineage kin
  Own sons                               21
  Daughters' sons                         5
  Sons' sons                              5
  Brothers' sons                          8
  Other kin                              37
Total II                                 76            26
III. Spouses and affines
  Husbands                                2            1

                                        B. FEMALE MEMBERS

                                       Number       Per Cent
IV. Lineage kin
  Full sisters                            9
  Full sisters' daughters                 7
  Daughters(*)                            9
  Sisters' daughters'
    daughters                             6
  Other matrilineal kin                   6
Total IV                                 37            13
V. Nonlineage kin
  Daughters                              17
  Daughters' daughters                    3
  Sons' daughters                         5
  Brothers' daughters                    13
  Other nonlineage kin                   41
Total V                                  79            27
VI. Spouses and affines
  Wives                                  34
  Wives' kin                              5
Total VI                                 39            13
Others                                    5            2
Male and female totals                   293          100

                                         A. MALE MEMBERS

                                            Strangers'
                                            compounds

                                       Number       Per Cent
I. Lineage kin
  Full brothers                           1
  Full sisters' sons                     ..
  Classificatory sisters' sons            1
  Sisters' daughters' sons                1
  Sons(*)                                 7
  Other matrilineal kin                   5
Total I                                  15            10
II. Nonlineage kin
  Own sons                               17
  Daughters' sons                        ..
  Sons' sons                             17
  Brothers' sons                          2
  Other kin                              16
Total II                                 52            34
III. Spouses and affines
  Husbands                                4            3

                                         B. FEMALE MEMBERS

                                       Number       Per Cent
IV. Lineage kin
  Full sisters                           ..
  Full sisters' daughters                 1
  Daughters(*)                            5
  Sisters' daughters'
    daughters                            ..
  Other matrilineal kin                   6
Total IV                                 12            8
V. Nonlineage kin
  Daughters                               7
  Daughters' daughters                    2
  Sons' daughters                         9
  Brothers' daughters                     4
  Other nonlineage kin                   19
Total V                                  41            27
VI. Spouses and affines
  Wives                                  20
  Wives' kin                              2
Total VI                                 22            14
Others                                    6            4
Male and female totals                   152          100

Source: Field Survey, Anglo and Suhienso, 1993 and 1996

(*) Female-headed households

Strategies of resource distribution differ according to household status. In the Akan world relationships of trust tend to follow the patrilineal line. Fathers have more trust in children than in nephews and potential successors. Holders of offices will therefore try to use their authority to have sons and daughters reside in their village (cf. Christensen 1954:57). Some chiefs manage to do so by carefully manipulating the norms which regulate access to land and power. Ahenemma who agree to reside in their fathers' village are offered protection, assistance, and a share of their fathers' prestige. Within the village, sons of chiefs are attributed roles of linguist and chief's councilor. They may also be called upon to hear court disputes. Daughters are often given a chance to start a trading activity (Fortes 1963:62; Mikell 1984; Palumbo 1992, 1994:117). The genealogies from Anglo and Suhienso (Figures 3 and 4) show that adehyee were successful in keeping their children's residence within their villages since the end of the nineteenth century. Since the 1940s the possibility of land access has become a key factor in an individual's decisions concerning residence. Apart from the advantages mentioned above, ahenemma were granted the right to clear the forest around their fathers' villages and thus acquire land titles.(15) By the 1960s, when the forest around the villages was getting exhausted and land access had become crucial, ahenemma were able to acquire rights over large tracts of land through the assistance and protection granted them by chiefs and mmusua panin. Normally one has the right to leave his private, self-acquired property to whomever he wishes, while the property inherited by members of the matrilineage should be transmitted matrilineally. Chiefs are supposed to leave part of their self-acquired property with the matrilineage, besides the stool land, the matrilineal land of the adehyee, which they acquired when they were enstooled as chiefs (cf. Danquah 1928:199-206). Chiefs and mmusua panin have repeatedly tried to evade these norms. They distributed part of their land, acquired as individual property, to their children and tried, with more or less success, to assign them part of the adehyee's matrilineal land.(16)

Ahenemma's households are thus the result of an original patrilineal transmission of wealth or grant of right from a holder of a matrilineal office to his child. The land of members of ahenemma's households is often individual property acquired through the first clearing of the forest or through gifts from patrilineal kin. The owner may therefore attribute the land to patrilineal or matrilineal kin during his lifetime (Table 3). Strategies of land transmission of adehyee's households follow a very different logic. Adehyee tend to recall members of their matrilineage who face problems of land access in other villages. They are offered land and are asked to settle in the matrilineal residential unit of the adehyee. By far the greatest share of plots obtained by adehyee is obtained from matrilineal kin. Land is transmitted matrilineally through succession from matrilineal kin or is granted as family land (abusua asase) by the abusua panin. The land thus obtained is considered family property and needs to be transmitted matrilineally. In adehyee's households, a matrilineal pattern of transmission is prevalent for adehyee. However, some young ahenemma are often present in adehyee's households while waiting for the conditions necessary to build their own, separate house. Chiefs and mmusua panin try to direct land toward their children, while adehyee need to check the attempts made by the heads of their matrilineage to privilege their children over their nephews (Table 3).

Table 3: Last Transmission of Plots of Land from Kin to Present Owner According to Kin and Status of Household

                                             Adehyee's
                                            compound(*)

                                       Adehyee        Non-adehyee

                                   No.          %     No.       %
I. Lineage kin
  Mothers' brothers                6                   ..
  Mothers                          13                    4
  Mothers' mothers' brothers       3                   ..
  Classificatory mothers'
   brothers                        12                  ..
  Family land                      3                   ..
  Other matrilineal kin            8                   ..
Total I                            45           84       4      16
II. Nonlineage kin
  Fathers                          2                    19
  Mothers' fathers                 1                   ..
  Other nonmatrilineal kin         1                   ..
Total II                           4             8      19      79
III. Spouses
  Wives                            1                   ..
  Husbands                         3                     1
Total III                          4             8       1       5
Totals I, II, and III              53          100      24     100

                                     Ahenemma's
                                      compounds

                                   No.          %
I. Lineage kin
  Mothers' brothers                60
  Mothers                          24
  Mothers' mothers' brothers       20
  Classificatory mothers'
   brothers                        2
  Family land                      3
  Other matrilineal kin            30
Total I                            139          63
II. Nonlineage kin
  Fathers                          56
  Mothers' fathers                 6
  Other nonmatrilineal kin         5
Total II                           67           30
III. Spouses
  Wives                            5
  Husbands                         10
Total III                          15            7
Totals I, II, and III              221         100

                                     Strangers'
                                     compounds

                                   No.          %
I. Lineage kin
  Mothers' brothers                ..
  Mothers                          ..
  Mothers' mothers' brothers       ..
  Classificatory mothers'
   brothers                        5
  Family land                      ..
  Other matrilineal kin            5
Total I                            10           30
II. Nonlineage kin
  Fathers                          22
  Mothers' fathers                 ..
  Other nonmatrilineal kin         ..
Total II                           22           67
III. Spouses
  Wives                            ..
  Husbands                         1
Total III                          1             3
Totals I, II, and III              33          100

Source: Field Survey, Anglo, 1996

(*) Members of the adehyee's compound were divided between adehyee and non-adehyee.

CONCLUSIONS

A first step toward the recognition of the influence of history on the pattern of residence has been made with the introduction of the historical status of households as a variable of quantitative analysis. Fortes's presentation of data ignores history in yet another respect. Data regarding the relationship between household members and household head imply an analysis confined to a generational period. The parental composition of households is studied only over the time span of two or three generations. A more satisfactory way of taking history into consideration is to analyze the relationship between present household members and the founder of the household. If the household's residential composition is considered with respect to the founder and not to the household head, the distinctive behaviors of adehyee's, ahenemma's, and strangers' households appear more clearly. Households founded by

The data help to shed light on the historical strategies of household composition. Adehyee's households have transmitted the position of household head solely through matrilineal lines. In ahenemma's and strangers' households, on the other hand, there were cases of patrilineal transmission of household headship. In these households some members who were classified as matrilineal kin of the household head (Table 2) are not matrilineal descendants of the household founder and therefore the degree of matrilineality in respect to the founder decreases with the flow of history and generations (Table 4). A similar device may be applied to the data concerning the transmission of land. In Table 3 the study of the transmission of land was limited to the last transferral of ownership. In the following table the transmission of land ownership is analyzed from the time the forest was cleared and farming rights over the land became individual property. The percentages of plots which have followed a wholly matrilineal transmission since the first clearing of the forest vary according to household rank. Adehyee who live in adehyee's residential units mostly acquired land which was first cleared by a member of their matrilineage and transmitted matrilineally since then. Non-adehyee (mostly ahenemma who live in the adehyee's compound), dwellers of households founded by ahenemma, and strangers received only a minority of their plots through matrilineal transmission from their kin who cleared the forest (Table 5).

Table 4: Classification of Members of Households in Relation to Household Founder and to Status of Household (Excluding Heads)

                               Adehyee's               Ahenemma's
                                compounds              compounds

                             Number   Per Cent    Number   Per Cent

Total lineage kin              79        62        74           25
Total nonlineage kin           28        22       173           59
Total spouses and affines      21        16        41           14
Total others                   ..        ..          5           2
Totals                        128       100       293          100

                               Strangers'
                               compounds

                              Number     Per Cent

Total lineage kin                15        10
Total nonlineage kin            105        69
Total spouses and affines        26        17
Total others                      6         4
Totals                          152       100

Source: Field Survey, Anglo and Suhienso, 1993 and 1996.

Table 5: Transmission of Plots of Land from First Owner to Present Owner According to Kin and Status of Household

                                          Adehyee's
                                          compound(*)

                                        Adehyee         Non-adehyee

                                      No.        %      No.       %
Transmission only through
matrilineal kin                         42       81       2       8
Transmission not exclusively
through matrilineal kin                 11       19      22      92
Totals                                  53      100      24     100

                                       Ahenemma's
                                       compounds

                                      No.        %
Transmission only through
matrilineal kin                        109       49
Transmission not exclusively
through matrilineal kin                112       51
Totals                                 221      100

                                        Strangers'
                                        compounds

                                       No.        %
Transmission only through
matrilineal kin                          9       27
Transmission not exclusively
through matrilineal kin                 24       73
Totals                                  33      100

Source: Field Survey, Anglo, 1996

(*) Members of the adehyee compound were divided between adehyee and non-adehyee.

A historical perspective has proven useful for two purposes. First, it defines the status of the social units and therefore considers them not as a constant but as a variable in quantitative analysis. Second, it reshapes numerical analysis to allow for the flux of people within households and for land ownership to be studied beyond the time span of a few generations. Attempts to analyze the Akan residential pattern according to models of evolutionary or generational time have omitted the effects of history on parental composition. A historical perspective, which shifts the notion of time from a merely generational and circular one to a historical and linear one, enables a distinction between genealogical origins, social behaviors, and the economic strategies of members of different residential units. The diachronic approach further suggests that patrilineal tendencies within a prevailing matrilineal system have been and are an essential structural component rather than an undermining and conflicting element.

NOTES

(1.) Fieldwork took place in 1993, 1995, and 1996. I thank the Italian NGO Ricerca e Cooperazione and in particular Dr. Gianna Da Re and Professor Antonino Colajanni for their support during fieldwork in 1993. My thesis at the University of Siena (Boni 1995), from which this article is drawn, was supervised by Dr. Luciano Li Causi and Professor Pier Giorgio Solinas, who contributed valuable comments. I am grateful to Karen Ludtke, Stephen Aidoo, and to the anonymous reviewers of Ethnology for their scrupulous reading of previous drafts of this article and for their much appreciated suggestions. The villagers of Anglo and Suhienso provided hospitality, advice, and friendship; and my sincere thanks go to my hosts and friends Stephen Aidoo, Oscar Nsowa, Felix Bib, and George Gyasi.

(2.) Fortes considers the prevalence of matrilineal kin of the household head in residential units in which the head is female "an important structural principle" (cf. Vercruijsse 1972:10-11). On the contrary, it appears to be a logical consequence of the fact that children of the household head do not belong to their father's matrilineage but share their mother's matrilineal membership.

(3.) Fortes's Table 4 considers only male-headed households. Female-headed households have a larger share of matrilineal kin (Fortes 1949:18, Table 3).

(4.) The word "queen mother" has been widely used in the literature that deals with the Akan but is misleading. The ohemma is not a queen, but the sister, niece, or the mother, real or classificatory, of the ohene. To put it simply, she is a female member of the matrilineage.

(5.) Some ancestors who predate the immigrant couple may be remembered, but the immigration marks a new beginning in the relationship between matrilineage and territory.

(6.) The idea of a direct coincidence between village, chief, and adehyee may be misleading. There are villages with more than one chief. In such cases, the land and the power may be divided between all or monopolized by one of them.

(7) Okali (1983) describes settlements of a similar kind. Her analysis is concerned with two mobile cocoa-farming communities with a high percentage of noncitizens. Okali (1983:31-44) introduces the variable of status of households in her study, but only to distinguish households headed by strangers and those headed by stool citizens. The two settlements are characterized by an inclusive residential pattern. The two communities she studied differ from most Akan settlements because in mobile settlements "the possibility of separating ties of descent, marriage and parenthood can no longer be maintained for residence purposes" (Okali 1983:44).

(8.) The high degree of independence allowed to the Anglohene is due to the royal families of Anglo and Nsuonsua considering themselves segments of one matrilineage. The Nsuonsuahene, who should be the one to administer the territory surrounding Anglo, agreed to leave some political autonomy to his "brother," the Anglohene.

(9.) Figures 3 and 4 (General Schematic Genealogy of the villages of Anglo and Suhienso) were elaborated considering the lines of filiation produced by the founders of the village. The lines of filiation were divided according to the residential units present in the village. This type of representation is aimed at showing the general dynamics of relations of filiation between kin-residential groups. The genealogical representation enables the inclusion of a large number of village residents and provides a diachronic picture of the dynamics that have produced the present residential pattern of the village. Such genealogical representation, however, has some limits. The two genealogies must therefore be read with great care and with an awareness of their representational flaws. More detailed genealogies have been included in a previous work (Boni 1995).

Some of those who were present in the village at the time of the field survey are not represented. The residential units which do not descend from the founders of the village (all stranger compounds) are not represented (with the exception of household number 7 of Anglo, linked through marriage to number 9). Only one line of filiation, either the maternal or the paternal, is represented. Those with both genitors who had kin groups within the village were represented only in the kin context of the residential unit where they resided. The numerous matrimonial ties between inhabitants of the village are normally omitted and so are most bonds of affinity.

Some members of kin groups which are represented have not been included in the genealogies. Members of the represented groups who were not present at the time of the survey (among others, migrants and women who had married outside the village) have been omitted. It was difficult to represent those who chose to live in a residential unit which was not the one in which either of their genitors resided; they were often excluded from the representation. Those who did not live in the village at the time of the survey but who had left some descendants in the village were included in the representation.

(10.) Normally the adehyee inhabit the palace (ahenefie) and the surrounding households in the center of the village. In the village of Anglo, the adehyee occupy one residential unit (household number 5); in the village of Suhienso they have six different households (numbers 1, 2, 4, 5, 10, 11).

(11.) This is a very frequent occurrence due to the high incidence of divorce and widowhood. For the high rate of divorce see Addo and Goody (1977:26), Fortes (1954:269), Oppong (1974:46), and Oppong and Bleek (1982:26-27). For the rate and effects of widowhood see Fortes (1954).

(12.) Kopytoff (1977) has shown the effects of this strategy on the geographical distribution of members of a Suku matrilineage in Zaire.

(13.) This is the case provided that the man is not a polygamist. If he is, the wives who do not belong to the royal matrilineage and their children may, at his death, claim part of the land.

(14.) Fortes's Table 4 (Table 1 in this article) considers only male-headed households, while I have included female-headed households which consist of four adehyee-headed and four ahenemma-headed households. While Fortes calculates separate percentages for male members and female ones, I have calculated joint percentages.

(15.) The right to clear the forest and acquire land ownership was granted to all Sefwi. Nonetheless, consent from the political authority on whose territory the forest was cleared was essential. Sefwi therefore acquired land rights principally around the village where their matrilineage or their father held rights over the territory (cf. Hill and McGlade 1957:3).

(16.) In the last decade the tendency to leave the major share of property to wives and sons has been strengthened by law (Awusabo-Asare 1990; cf. Woodman 1974:268-84).

BIBLIOGRAPHY

Abu, K. 1983. The Separateness of Spouses: Conjugal Resources in an Ashanti Town. Female and Male in West Africa, ed. C. Oppong, pp. 156-68. London.

Addo, N. O., and J. R. Goody. 1977. Siblings in Ghana. University of Ghana Population Studies No. 7. Legon.

Akwabi-Ameyaw, K. 1982. Ashanti Social Organization: Some Ethnographic Clarifications. Ethnology 21(4):325-33.

Amoo, J. W. A. 1946. The Effects of Western Influence on Akan Marriage. Africa 16:228-37.

Arhin, K. 1983a. Peasants in Nineteenth Century Asante. Current Anthropology 24(4):471-80.

-- 1983b. Rank and Class among the Asante and Fante in the Nineteenth Century. Africa 53(1):2-22.

-- 1986. The Expansion of Cocoa Production in Ghana: The Working Conditions of Migrant Cocoa Farmers in the Central and Western Regions. Legon.

Awusabo-Asare, K. 1990. Matriliny and the New Intestate Succession Law of Ghana. Canadian Journal of African Studies 24(1):1-16.

Basehart, H. W. 1961. Ashanti. Matrilineal Kinship, eds. D. Schneider and K. Gough, pp. 270-97. Berkeley.

Benneh, G. 1970. The Impact of Cocoa Cultivation on the Traditional Land Tenure System of the Akan of Ghana. Ghana Journal of Sociology 6(1):43-61.

Bleek, W. 1972. Geographic Mobility and Conjugal Residence in a Kwahu Lineage. Institute of African Studies Research Review 8:47-55.

-- 1975. Parents and Children in a Kwahu Lineage. Legon Family Research Paper 4:30-38. Legon.

Boni, S. 1995. Parentela e insediamento tra i Sefwi del Ghana: Le relazioni tra politica, residenza e rapporti parentali in un'etnia Akan. B.A. thesis, Universita di Siena.

-- 1997. Storia e parentela nella creazione e affermazione dell'Oman di Sefwi Wiawso (c. 1650-1887). Africa (Roma) LII(1):28-61.

Christensen, J. B. 1954. Double Descent among the Fanti. New Haven.

Danquah, J. B. 1928. Akan Laws and Customs. London.

Fortes, M. 1949. Time and Social Structure: An Ashanti Case Study. Time and Social Structure and Other Essays, M. Fortes, pp. 1-32. New York.

-- 1950. Kinship and Marriage among the Ashanti. African Systems of Kinship and Marriage, eds. A. R. Radcliffe-Brown and D. Forde, pp. 252-84. London.

-- 1953. The Structure of Unilineal Descent Groups. Time and Social Structure and Other Essays, pp. 67-95. New York.

-- 1954. A Demographic Field Study in Ashanti. Culture and Human Fertility, ed. F. Lorimer, pp. 253-339. Paris.

-- 1958. Introduction. The Developmental Cycle in Domestic Groups, ed. J. Goody, pp. 1-14. London.

-- 1959. Descent, Filiation and Affinity. Time and Social Structure and Other Essays, pp. 96-121. New York.

-- 1963. The "Submerged Descent Line" in Ashanti. Studies in Kinship and Marriage, ed. I. Schapera, pp. 58-67. London.

-- 1970. Kinship and the Social Order: The Legacy of Lewis Henry Morgan. London.

-- 1971. The Family: Bane or Blessing? Accra.

-- 1974a. The Prologue: Family Studies in Ghana 1920-70. Domestic Rights and Duties in Southern Ghana, ed. C. Oppong. Legon Family Research Paper 1:1-27. Legon.

-- 1974b. The Akan Family System Today. Domestic Rights and Duties in Southern Ghana, ed. C. Oppong. Legon Family Research Paper 1:28-34. Legon.

Fortes, M., R. W. Steel, and P. Ady. 1947. Ashanti Survey 1945-6: An Experiment in Social Research. The Geographical Journal CX(4-6):149-79.

Gough, K. 1961. The Modern Disintegration of Matrilineal Descent Groups. Matrilineal Kinship, eds, D. Schneider and K. Gough, pp. 631-52. Berkeley.

Hardiman, M. 1974. A Preliminary Study of the Role of Women in Some Akan Rural Communities. Domestic Rights and Duties in Southern Ghana, ed. C. Oppong. Legon Family Research Paper 1:102-28. Legon.

Hill, P. 1975. The West African Farming Household. Changing Social Structure in Ghana, ed. J. Goody, pp. 119-36. London.

Hill, P., and C. McGlade. 1957. An Economic Survey of Cocoa Farmers in Sefwi-Wiawso. Cocoa Research Series No. 2. Legon.

Kobben, A. J. 1954. L'Heritage chez les Agni: L'Influence de l'econimie de profit. Africa 24:359-63.

Kopytoff, I. 1977. Matrilineality, Residence and Residential Zones. American Ethnologist 4(3):539-76.

Kyei, T. E. 1992. Marriage and Divorce among the Asante: A Study Undertaken in the Course of the Ashanti Social Survey (1945). Cambridge.

Lystad, R. A. 1959. Marriage and Kinship among the Ashanti and the Agni: A Study of Differential Acculturation. Continuity and Change in African Cultures, eds. W. R. Bascom and M. J. Herskovits, pp. 187-204. Chicago.

McCaskie, T. C. 1980. Office, Land and Subjects in the History of the Manwere Fekuo of Kumase: An Essay in the Political Economy of the Asante States. Journal of African History 21:189-208.

-- 1995. State and Society in Pre-colonial Asante. Cambridge.

Mikell, G. 1984. Filiation, Economic Crisis, and the Status of Women in Rural Ghana. Canadian Journal of African Studies 18(1): 195-219.

Okali, C. 1983. Cocoa and Kinship in Ghana: The Matrilineal Akan of Ghana. London.

Oppong, C. 1974. Marriage among a Matrilineal Elite--A Family Study of Ghanaian Senior Civil Servants. Cambridge.

Oppong, C., and W. Bleek. 1982. Economic Models and Having Children: Some Evidence from Kwahu, Ghana. Africa 52(4):15-33.

Palumbo, B. 1991. Le noci della discordia: Terra, eredita e parentela in una comunita Nzema (Ghana). Ph.D. dissertation, Universita degli Studi "La Sapienza." Rome.

-- 1992. Marriage, Land and Kinship in a Nzema Village. Ethnology 31(3):233-57.

--1994. Storia, spazio e potere in un villaggio Nzema. Etnosistemi 1(1):102-26.

Pavanello, M. 1995. The Work of the Ancestors and the Profit of Living: Some Nzema Economic Ideas. Africa 65(1):36-57.

Perrot, C.-H. 1982a. Les Anyi--Ndenye et le pouvoir aux 18e et 19e siecles. Paris.

-- 1982b. Formation d'etats et formation d'une ethnie: Le Cas des Anyi-Ndenye. Cahiers d'Etudes Africanes XXII(3-4):455-63.

-- 1985. L'Appropriation de l'espace, un enjeu politique. Pour une historie du peuplement. Annales Economies Societes Civilisations 40(6): 1289-1306.

Rattray, R. S. 1927. Religion and Art in Ashanti. London.

-- 1929. Ashanti Law and Constitution. New York.

Signorini, I. 1977. Il modello di residenza. Una societa guineana: Gli Nzema, ed. V. L. Grattannelli, pp. 291-320. Turin.

Tellier, M. 1902. Coutumes des Agni de l'indenie. Les Coutumes indigenes de la Cote d'Ivoire, eds. F. J. Clozel and R. Villamur, pp. 147-70. Paris.

Terray, E. 1976. Contribution a une etude de l'armee Asante. Cahiers d'Etudes Africanes XVI(1-2): 297- 356.

-- 1987. Le Royaume Abron du Gyaman. Princes et serviteurs du royaume, ed. C. Tardits, pp. 24-57. Paris.

Tetteh, P. A. 1967. Marriage, Family and Household. A Study of Contemporary Ghana, vol. 2, eds. W. Birmingham, I. Neustadt, and E. N. Omaboe, pp. 201-16. London.

Vercruijsse, E. 1972. The Dynamics of Fanti Domestic Organization: A Comparison with Fortes's Ashanti Survey. The Hague.

-- 1974. Composition of Households in Some Fanti Communities: A Study of the Framework of Social Integration. Domestic Rights and Duties in Southern Ghana, ed. C. Oppong. Legon Family Research Paper 1:35-56. Legon.

Wilks, I. 1967. Ashanti Government. West African Kingdoms of the Nineteenth Century, eds. D. Forde and P. Kaberry, pp. 206-38. London.

Woodford-Berger, P. 1981. Women in Houses: The Organization of Residence and Work in Rural Ghana. Antropologiska Studier (Stockholm) 30-31:3-35.

Woodman, G. 1974. The Rights of Wives, Sons and Daughters in the Estates of Their Deceased Husbands and Fathers. Legon Family Research Paper 1:268-84. Legon.

Zajaczkowski, A. 1960. La Famille, le lineage et la communaute villageoise chez les Ashanti de la periode de transition. Cahiers d'Etudes Africanes 1(4):99-114.