GEOGRAPHICAL MOBILITY AND CONJUGAL RESIDENCE IN A KWAHU LINEAGE * By Wolf Bleek Students of the Akan social system and matrilineage systems in general have for some time been engaged in a discussion as to whether the system is "breaking down" or not. Terms such as "disintegration" (Gough, 1961) and "Westernization" (Caldwell, 1968; Jahoda, 1961-2) have been introduced and rejected again for several reasons. Both sides in the debate point to the same phenomena to prove their thesis, stressing opposite aspects. P\ules of succession, domestic organization and conjugal relationships are among issues most frequently discussed. The "heart of the matrilineal matter" has been defined by Audrey Ri- chards some years ago. She has termed it the "Matrilineal Puzzle" by which she means, "the difficulty of combining recognition of descent through the woman with the rule of exogamous marriage." (Richards, 1950: 246). Autho- rity over the wife, therefore, has to be shared between husband and wife's brother, which is very "puzzling" indeed and easily results in a tense situation. The crucial factor in this"tug-of-war" is, according to Richards, "the question of residence at marriage." (1950: 249) (emphasis ours). The one with whom the wife is staying is likely to have the most authority over her. Accepting her definition, it looks tempting to take conjugal residence as an index of the strength of matrilineality. Virilocal+ residence could be taken to signify that matrilineal ties are weak and have given way to conjugal ties; duoloeal residence, on the other hand, would mean that matrilineal pulls still dominate over the conjugal ones. An increase of virilocal residence at marriage could therefore be interpreted as a loss in strength of the matri- lineage and an increasing tendency towards the conjugal family. * Took M.A. at the Department of Sociology, Legon and subsequently carried out further fieldwork on Akan social structure under the auspices of the Institute of African Studies. +"Virilocal" means here: residence of the wife with the husband irrespective of wihether he has his own house or resides with his matrikin. 48 As a matter of fact, virilocal residence seems on the increase among the Akan. The first to notice this was Fortes who compared two Asante towns and found that the one which was most exposed to "Western" influence had much higher percentage of wives residing with their husbands (Fortes, 1949). A similar discovery was made by us in a depth-study of a Kwahu matrilineage (van der Geest, 1972). In this lineage conjugal residence has developed from duolocal to dorainantly virilocal during the past 80 years. A closer examina- tion of the data revealed, however, that this fact could not unequivocally be attributed to a "decrease in matrilineality", but that the predictor variable was an entirely different factor. In this exercise we will be concerned with one section ofthe matrilineage under study. This section insludes 41 married members spread over 5 genera- tions and extending over a period of approximately 80 years. In total we recorded 132 marital unions among them, most of which ended in divorce or separation and some in death of one ofthe partners. A few are still in existence The 132 unions comprise both customary marriages and, what we have called "Free Marriage" (in Twi: mpena awaree), a looser type of union which is es- pecially popular among the most recently married generation of this lineage. The town in which we carried out our research is situated on the Kwahu Plateau and has about 5,000 inhabitans according to the 1970 census. It is a typical rural Ghanaian town, occupied mainly by children and elderly people. Employment opportunities for school leavers and others who do not want to go into farming are minimal and most of them have left for Accra or other places. The rest of the middle group — those who choose farming — are staying in town, but more often than not, they moved to their farming villages, 10 to 30 miles away. Marital residence, we found, is closely related to this migratory pattern. Couples living in the farming villages, in the bigger towns, or in fact, anywhere outside their home town tend to stay together in one house (or room). Couples who live in their home town, however, do not usually reside to- gether. This trend holds true, independent of education, church adherence, status, seniority of marriage, number of children and other factors which are some times brought to the fore. One factor, however, should be added to the above one, the reason for which can be readily understood, that is, whether the husband and wife are from the same town. If they are not, e.g., if the wife had to leave her home town after her marriage, residence will also tend to be virilocal. Adding the two factors together — migration and wife's origin — we can say that the crucial factor determining type of residence is whether the couple is staying in the wife's home town or not. In the table below, therefore, we have taken this factor as the independent variable. •4 A m 1 4 4 Table I: Town of residence and conjugal residence. 49 Residence together Separate residence Un-known Total Couple staying in wife's home town 18 52 Couple staying outside wife's home town Unknown Totals 54 5 77 6 1 59 6 1 4 II 76 61 10 147* df = I x2 = 55.22 p <0.00l * This total is higher than the total of 132 marriages because some marriages show dif- ferent residence patterns at different stages. The implications of Table I are clear. As long as the wife's matrilineage is around, she tends to reside with them and not with the husband. If her matrikin are far away, i.e. in another town, she will take up residence with her husband. The correlation in Table I is extremely significant in statistical terms (P <0.00l), butit should be borne in mind that the sample of 130 (the 17 unknown cases have been omitted) is rather small and can hardly be con- sidered representative of the Kwahu, let alone the Akan, since all the units have been selected from one single matrilineage. Let us now have a closer look at how stringently these rules apply. Female members who were asked whether they stayed with their husband in the same house, frequently gave as the explicit reason why they did not, "Because my 'family house' was there", or, if they did stay together, "Be- cause there was no one of my family I could stay with". Male members gave similar replies with respect to their wives. When a couple moved from the wife's home town to elsewhere or vice versa, this was usually accompanied by a change in residence pattern. To give a few examples, one man was married to two wives about 35 years ago, both from his home town. One he took to Swedru where he was trad- Ing and she stayed in the same house with him there. But whenever they visited their home town, the wife went to stay with her own relatives, and k. 50 she finally settled again in her "family house" when the man gave up trad- ing and returned home. The other wife he left in his (and her) home town while he was trading at Swedru and she resided with her matrikin. When he stopped trading and went into cocoa farming, he took this wife along to his cocoa village where they stayed together in one house. But during visits to their home town they again separated, he stayed with his relatives and she with hers. Similar examples can be cited for the present time. By the same token, if husband and wife hail from different towns they do not separate while visiting their home towns, e.g. a female member married to someone from another Kwahu town, lives with her husband and their children in Brong Ahafo. When they come for a visit to his town, she stays in his "family house." Again many more examples could be cited. As a conclusion we can say that, when conjugal and matrilineal ties have "equal chances," i.e., when the couple live in the wife's home town, matri- lineal ties are considerably stronger than conjugal ones. The type of residence at marriage, therefore, which prima facie, looked a useful tool to detect change in the matrlliny proves to be very deceptive in this respect. Type of residence is, rather, related and — as we have demonstrated — casually connected, with geographical mobility. Increase of geographical mobility results in an increase of virilocal residence, as couples move away from the sphere of influence ©f their matrikin and more husbands find their wives outside their own home town. It is probably also in this way that we must interpret the findings of Fortes at Asokore and Agogo (Fortes, 1949): Asokore, situated on the mainroad to Kumasi, must have had a higher degree of geo- graphical mobility than Agogo. As a result more Asokore men must have had their wives selected from outside Asokore. This, however, is only a hypo- thesis; information to test it is lacking. Our hypothesis does not deny that, ultimately, geographical mobility — which is related to many other factors that lie outside the scope of this paper — can have a strong effect upon the matrilineal system itself. It pro- bably will. For the present, however, there are no indications that, indepen- dent of geographical mobility, there is any rise in the incidence of common residence at marriage. To demonstrate this, we have to look at the incidence of common residence while staying in the wife's home town over a certan period of time. We have for that purpose broken down Table I into generations (see Table 2). 4 -A i A - • ** -•4 4 A t 4 51 * it Table 2: Town of residence and conjugal residence according to generation. Generation 1 II III IV+V Residence tog dl tog dl tog dl tog dl ? Totals In Wife's home town Not in wife's home town Unknown Totals 0 0 0 0 8 0 0 8 7 19 75 15 4 10 6 78 20 0 5 0 18 5 1 1 16 0 0 0 1 4 59 10 27 24 30 17 20 10 II 147 Legend: tog = together; dl = duolocal The table shows in the first place a marked increase of common residence corresponding with a less marked increase of geographical mobility. The unknown cases are slightly disturbing here. Information we gained about the first generation was rather vague and equivocal and should be treated with caution as errors may well have been made. Moreover, the sample of eight for that generation is too small to enable us to draw any firm conclusion. Let us now concentrate on the category "common residence in the wife's home town." It scores 7, 7 and 4* for the second, third and combined fourth and fifth generation, which is respectively 37%, 4 1% and 40% of the total of marriage in the wife's home town in those generations. On the basis of these scanty data the increase of common residence at marriage cannot be in- terpreted as a shift in emphasis from matrilineal to conjugal ties, but only as a necessary result of a growing geographical mobility. This — as * Out of these four, two are cases of uxorilocal residence in which the husband was a stranger in the wife's home town. 52 stands to reason — applies only to the matrilineage under study. A much larger and more stratified sample is needed to prove whether these con- clusions have a wider applicability. Finally, we should take a brief glance at some other factors which have been, or are likely to be, presented as attributing factors in common resi- dence at marriage. It was found that "Free Marriage" (mpenaawaree) which is a more tem- porary type of union, does not necessarily imply duobcal residence as might perhaps be expected. On the contrary, as "Free Marriage" is often contracted away from home, it is characterized by geographical distance and, conse- quently, by common residence. Some scholars have suggested that type of residence may reflect a stage in one's life-cycle, rather than a stage in the process of social change of a society: Fortes has, long ago, suggested that "Residence patterns are the cry- stallization, at a given time, of the developmental process" (1949:3). How this developmental process evolves in the case of the Asante he describes in the same study. "During the first two or three years of wifehood the great majority of young wives continue to reside with their own kin... As they advance in maturity the pull of conjugal ties increases and reaches its maxi- mum at the peak of the child-bearing years, in the thirties, By this time, they have perhaps three children, for whom it is an advantage to be under their father's care. Finally, when the child-bearing years are over and their child- ren are grown up, the desire to establish their own households becomes strong in many women, the more so if they are widowed or divorced." (Fortes, 1949:78). Fortes corraborates his statement with a table (1949:77) showing that the percentage of wives staying with their husbands is highest between the ages of 20 and 40. His figures are, however, deceptive as they are percentage* of all women in that age category, both married and unmarried. Unmarried women are, however, considerably more numerous outside the 20-40 age category, and it is therefore not surprising that the absolute number of women staying with a husband outside the 20-40 age category is much lower. But if we take the figures in the table as percentages of all the married women in that age category, we see no longer a "beautiful line" of development. At Agogo, for example, the number of wives in the 16-20 age category staying with their husband is even slightly higher than those in the 21-25 and 26-30 category, and at both Asokore and Agogo the average number of wives above 40 years staying with their husband is higher than the average number of those below 40. Fortes' statement is, therefore, refuted by his own data, and is, probably, mainly supported by what people think happens or should happen. Old people in the town of our research expressed the same view, i.e., when a man grows older and he and his wife have many children, he is expected to build his own house to accommodate his wife and children. But, as Fortes. ' • %• -i •a £ S Jfc k 53 himself observes, this ideal is less often realised than one would believe (1949:69). Our diachronic investigations do not provide a reliable division into age groups to bring out the extent to which this ideal has been realised in our lineage at issue. What we can do, however, fs to record the order in which common and separate residence alternate in the life-cycles of the 41 members of the lineage. Change in conjugal residence, it should be remarked, may happen within one marriage and is then usually the result of the couple moving from one place to another. More common, however, is a change in residence to coincide with a change of marriage partner, either after death or divorce of the former partner. In the table below we have treated the unknown cases as non-existent, which, unfortunately, distorts the picture to some extent and eliminates one member from the list. We are not concerned with the length of any residence type, nor whether it refers to one or more unions. So, the first category (s) can refer to one life-long marriage during which the partners continued to reside separately as well as to a series of subsequent marriages which were all characterized by separate residence. Table 3: Conjugal residence cycles in a Kwahu lineage s s-t s-t-s t t-s t-s-t more complex Total 8 9 0 12 4 1 6 40 Legend: s — separate; t — together The table does not bring out a particular order as most probable, and we are not surprised to find that the three highest scores (12, 9 and 8) follow 54 roughly the migratory pattern of the people concerned. The 12 members who stayed together with their partner(s) throughout their life prove to be the ones who were most away from their home town and the 8 who never stayed with their partner(s) were the ones who travelled least. Five of the 9 who changed from separate to together moved out of their town at a later stage in their life. A last factor worth examining in relation to conjugal residence is number of children. Both Fortes and elderly people in the town of research state that the growth of the number of children may instigate a man to build his own house and take his wife with him. As Fortes mentions the number "three" we shall take the number of three children as the borderline between the two groups of couples. We have attempted to include all live birth, also those children who died at a fairly young age, but it may well be that some children who died very early were not remembered as members of the first two generations. The fact that marriages with less than three children outnumber the other ones so greatly is mainly due to the high divorce rate: many marria- ges were terminated before they could produce three children. Table 4: Conjugal residence and number of children. (Percentages in brackets) Residence together Separate residence Unknown Total 53 (49) 47 (43) 9(8) 109(100) 24 (63) 12 (32) 2(5) 38 (100) Couples with less than 3 children Couples with 3 or more children Totals 77(52) 59 (40) II (8) 147 (100) According to Table 4 couples with three or more children are more likely to stay together than the ones with less than three children, but the correlation is not statistically significant. Moreover the sample is too smalf and response errors are probably too numerous to suit any significant test. Elaboration may well reveal partial associations which render the association "number of children — type of residence" spurious. In conclusion, we can say that in the lineage under study residence away from the wife's home town can predict common residence at marriage with a high degree of probability. Increase of this type of residence does not in- dicate that marriage is being 'Westernized,' nor that the matrilineal descent group is "disintegrating'-; it merely-means that geographical mobility is in- •4* * 55 creasing in the lineage concerned. Matrilineal ties, as this exercise has de- monstrated, continue to dominate conjugal ties where the two have "equal chances". It would be interesting to pursue this correlation further over a wider population of Akan, including a proportion of the elite who are frequently believed to hold more 'Western' ideas about marriage and family organization. One could ask whether the individuation of the elite conjugal family is a consciously pursued goal, or whether it is a mere adaptation to new economic and demographic situations as Oppong has suggested (1970: 306). References: Caldwell, J. C. — 1968: Population growth and family change in Africa; the new urban elite in Ghana. Canberra, Australian National University Press. Fortes, M. — 1949: "Time and social structure: an Ashanti case study." in: M. Fortes (ed) (1949) Social structure; studies presented to A. R. Radcliffe-Brown. Oxford Clarendon press, 54-84. Gough, Kathleen — 1 9 6 1: "The modern disintegration of matrilineal descent group" in: D. Sch- neider and Kathleen Gough (eds) (1961) Matrilineal Kinship. Berkely, University of California Press. 631-652 Jahoda, G. — 1961-62: "Aspects of Westernization. A study of adult class students in Ghana." in: The British Journal of Sociology, XII, 375 - 86 and XIII, 43-56. Oppong, Christine — 1970: "Aspects of conjugal relationships among Akan senior civil servants in Accra". Unpublished Ph.D. Thesis, University of Cambridge. Richards, Audrey — 1950: "Some types of family structure amongst the Central Bantu", in: Rad- cliffe-Brown and Fortes (ed) (1950) African systems of kinship and marriage. London, O.U.P. 207-51 van der Geest, S. — 1972: "Family dynamics in a changing society; a case study of a rural Ghanaian community". Unpublished M. A. Thesis, University of Ghana.