INHERITANCE, PROPERTY, AND MARRIAGE IN AFRICA AND EURASIA
111
Inheritance, Property, and Marriage in Africa and Eurasia
Jack Goody
Many of the de^kd_ob^rratiom of jjre-colpnlal.. Africaa..m West Africa, and especially from the Gold Coast. For this region was not only the closest part of Black Africa but also of the greatest economic interest, especially when the Portuguese lost their monopolistic hold on the coast and the way was open for the Protestant business men of the western seaboard, Holland, England, Scandinavia, and north Germany to develop the interlocking trade in gold, slaves, and firearms which stood them in such stead in the early years of capital accumulation that preceded the development of industry in Europe.
During this period, at the end of the seventeenth century, the Dutch factor, William Bos-man, was struck by certain features of social organization of the Gold Coast which he saw as fundamentally different from those he hacl grown up with in Western Europe. JSpjsttjari was on the coast for some 14 years and published hisjo^ervations in 1704 finder the title of A New and Accurate Description of the Coast of Guinea, divided into the Gold, the Slave, and the Ivory Coasts. In the twelfth letter of his book, the author writes 'Of the Negroes manner of Marrying'. 'Marriage here is not over-loaded with Ceremonies, nor have tney"any Notion of a Praevious Courtship to bring on a Match: here are no tedious Disputes
on account of Marriage Settlements____The
Bride b_rings.„.ttQ„1Qthgr. fipdy, nor does the Man want much; 'tis sufficient if
he has enough to^defray the Exjjence_fj£j^_e Wedding^Day' (l%"77l97-98)7 He further observes the corollary of the absence of do that 'Married people here have no community of Goods; but each hath b|s_or her particular Propriety.... On the Death of either the Man oFtKeWife, the respective Relations come and immediately sweep away all, not leaving the Widow or Widower the least part thereof...1 (202).1
Thus Bosman sees that not only is conjugal./;-1 community of property rare, but that a^ecea- • sect's estate is not called upoji.tp_supportjdie .= . surviving spouse. This fact is linked to the absence of a marriage settlement, to die absence of a woman's portion of the patrimony which she brings with her into marriage as a dowry.2 \ If a woman brings.nothing at marriage, she gets nothing when the union is dissolved. Bosnian, also notes a related fact, though he does not perceive its interrelatedness. He otjseryesjhat except at Accra, inheritance is imatriHneaj^ Even in the matrilineal societies property_js.. sex^inF^.Tr'tK'e'el'dest Son of his Mother is"*\ Heir to his Mothers Brother or her Son, as the eldest Daughter is Heiress of her Mothers Sister or her Daughter' (203). This is to say, property descends 'homogeneously' e.g. between males, even when it goes through females.3
In earlier publications, I have emphasized the importance of inheritance as a variable (1958, 1959, 1962) and suggested that in the domestic domain one of the major differences between
• Africa" and Eurasian societies lies in the fact that in Eura^^iy^rging.inheritance (i.e. 'bilat-eral' inheritance^jwhere property goes to .children of both sexes) is common, especially in the ^Jor civifizatr6ns,"'v^eHasm Africa it is virtu-ajjyjmJmown; The absence of diverging inheritance is linked to the absence of dowry in Africa (cf. Goode 1963: 67), since dowry is essentially 3 process whereby parental property is distributed to a daughter at her marriage (i.e. inter vivos) rather than at the holder's death (mortis causa). I therefore include dowry as part of the process of 'diverging devolution'. The property a woman receives through dowry or^bilateral' inhentanceestablishes some variety^pf. a.£pnju-gal fund, the nature of which may vary widely. .. ffiis fund ensures her support (or endowment) in widowhood and eventually goes to provide fofher sons and daughters. I have elsewhere tried to analyse the con-. comitants of diverging devolution in the Eurasian setting and have discussed some of the implications of this difference for comparative sociological studies. In a paper on 'Adoption in Cross-Cultural Perspective'  (1968), I have outlined some of the interlinking variables (unavoidably these are mostly 'qualitative variables" in the statistical sense) that I regard as concomitants of diverging devolution. I should add that I see the system of the transjiussion of property (i.e. devolution) as being the mde-pmdeht 'variabTe^m some of these instances and as the intervening_or_dependent variable in others; for in the main it seems to occur where (agnciHture^is intensive and the means of pro-auction are in relatively short supply. In this present paper I try to test certain hypotheses concerning diverging devolution by another means, namely, the recent Ethnographic Atlas. The 1967 version of this sample consists of 80 columns of data on 863 societies from all parts of the globe. This instrument enables us to check some of the statements that sociologists generate about societies, usually on the basis of the one or two in which they have worked and the handful they currently have in memory store. Clearly the number of aspects that one can test is limited. For example, it could be suggested that the extent of ceremonial performed at rites of passage in the individual's life cycle is positively correlated with the amount of work (in terms of the handing over
of rights and duties, etc.) that has to be done. Where marriage establishes a conjugal fund (as in dowry systems) the wedding ceremonial will be more elaborate than where it does not; where funerals redistribute the dead man's property, they will be more elaborate than where a holder divests himself of his property during his lifetime (Goody 1962: v). But this information is not coded in the present Ethnographic Atlas and to read through a representative sample of the relevant material would be very time-consuming. In other cases the information exists, though not always m quite the form one optimally needs. When this is so, I have carried out tests of the relevant aspects of the hypothesis.4 In so doing, I am not seeking to explain all of one 'variable' by another; in the example I gave of the relationship between life-cycle ceremonies and devolution, it is obvious that other factors are at work. The predictions are for a positive association, a significant trend, not a one to one relationship. In many branches of the social sciences a hypothesis is rejected if the probability value is less than .05, while an association of 0.33 on a phi test is understood to be high (the scale runs between -1 and +1); to this convention I will adhere in this paper.5
There are two kinds of problem involved in trying to make such a test, namely, those to do with the instrument and those to do with the analytic concepts. From the analytic standpoint, it is the transmission of major items of property that is clearly going to be of greater significance, especially the transmission of basic productive resources (usually land); but in the code the distinction between land and movables is made for inheritance but not for dowry. Secondly, there is a potential difficulty in deciding when property diverges. For example, a daughter may inherit her father's property in her own right or as an epiclerate, that is, a residual heir in the absence of brothers. Nevertheless the overall distinction is clear. In the main Jiurasian societies, a close female inherits before a more distanfmale, even where both are members of the same patrilineal descent group. In^fxica..AQHtb,„pf,^the^Sahara, a woman only inherits male property when there are no males left in the wider kingroup, and even then it is a very rare occurrence.
The specific problems to do with the instrument are twofold. Firstly the composers note
112
JACK GOODY
4
that the inheritance data have not been easy to code; indeed they describe the code as inadequate (Murdock 1967: 167). The second point has to do with marriage transactions. Since I define devolution as transmission between holder and heir (see Goody 1962:312), whether or not it takes place at death, I include dowry in these operations. Indeed I include not only the 'direct \3owry^ (the property passed from 'parents' to a daughter on her marriage)'. I also include the (indirect dow.ry',]that is, property passedj?y_the groom to the bride at marriage^Such prestations are often spoken of asyarjueweait^^^ but I would limit these terms to prestations that pass between the kin of the groom and the kin of the bride, and that can therefore be used to provide wives for the girl's brothers; in short, they formjjart of a system of circulating or on-going exchange. I suspect that most accounts fail to make a distinction between these types of prestation, despite the different social implications that they have. Hence what I would regard as (indirect) dowry may sometimes have been reported as brideprice.
As with most kinds of sociological analysis the measuring rods and the measurements themselves are bound to be less than perfect. But however crude, even such rough comparisons provide some degree of confirmation or contradiction of hypotheses about human social organization. On the one hand this procedure gets us out of the unsystematic comparisons upon which so much comparative sociology is based and on the other hand liberates us from certain of the limitations of the structural-functional method. It is no part of my intention to substitute one approach for another; different methods answer different problems. Indeed it is a sign of the relative immaturity of the social sciences that so many of their practitioners presume that there is a single approach to the sociological verities, an attitude which makes them more akin to philosophers than to other behavioural scientists.
Bearing in mind the limitations of this and other methods, I first looked at the distribution of diye£gi£JUk^
erty complex')6 in different continents, that is, thVHIstribution of those systems where diverging inheritance oridowry are found. This information was found in the following form in the Ethnographic Atlas:
ji) all societies where daughters have a share in either land (column 74: c, d) or movable property (76: c, d); (ii) all societies with dowry as the main or alternative method of marriage transaction (12: d; 13: d). The negative cases consist of all those that remained once the diverging devolution data had been extracted, less those societies with no individual property rights or no rule of transmission (74: o; 76: o) and those societies where there was no information on the relevant columns (74, 76, 12).
The first run was a purely distributional one, in order to confirm or refute the suggested differences between Africa and Eurasia. The following points emerge from it: ((ij) In Africa, diverging dev^^^
i.e. it occurs in _5j^r cent of the cases. Bosnian's observation thus has general' significance, ((ii), In America, there is a large proportion of societies with no individual.property rights or transmission rule. The reason is that in many of the hunting and gathering societies individuals had little property, except personal equipment, and this was often destroyed at death.7 On-going productive property was minimal.
i,(iii]) In America, the relatively large jiumber of societies with diverging inheritance is in part a question of the importation of European norms through imperial conquest. The Mayan-speaking people of Yucatan have the same practice (Redfield 1934: 61 ff.), but I do not know whether this is an aspect of early Mayan inheritance or a European import (the people of Chan Kom have long been Christian). Eurasian religious and secular codes (including Islam) promote diverging devolution independently of other factors, (iv) In Eurasia and the Pacific, the number of societies with diverging devolution is approximately 40 per cent of the total. However, this figure includes the major civilizations in the area, whose populations are more numerous and whose influence greater. It is suggested thatjjhe spcie.tie_s_with homogeneous devolution are. mostly those outside the majorjtrad-
1NHERITANCE, PROPERTY, AND MARRIAGE IN AFRICA AND EURASIA
113
]  itions^i-e. 'tribaP jioc^t^e.s. of,..various Hnds^pe^iajiy those ^without intensive agriculture.! In ^nij^riie^djstnbutional table shows that systems of devolution in Africa differ 'From those associated"witrTthe major Eurasian civil-ff izations, being of the homogeneous rather than {✓-the diverging Jcinci.
What about the correlates of these differences in devolution? It seems probable that where women are receivers of tna|ejp_roperty considerable j^termon^ their^ marriage arrangements. An heiress is often not allowed to marry just anyone; her spouse is more likely to be chosen for her. Other women too are likely to marry (and to want to marry) within rather than without; for unless a woman marries 'well' her standard of living might drop, and this would be a matter of concern to herself as well as her kinsfolk. It seemed plausible to test this assumption by means of the data on pre-marital virginity (col. 78); a stress on virginity at mar-nagTcoiild be held to indicate, inter alia, the degree of control exercised on women by society, kin, and self. It also limits the possibility of conflicting claims on the estate in which a woman has rights.
A further hypothesis is generated by this suggestion that where women^re jecipients of male property, there will be a greater tendency t^COTtroI~tfieir marriages. When women are heirs, or even residual heirs, they may be encouraged to marry within a certain range of kin; this was the case with the daughters of Zelophehad in ancient Israel as well as in the epiclerate of classical Athens. The Atlas does not permit us to assess the incidence of these kinds of marriages but it does enable us to get an idea of the association of one form of in-marriage, that is, endogamy (in kin, caste, or local group), with diverging devolution____Defined •culturally', in terms of Hindu ideology, caste is clearly limited to the Indian subcontinent. Defined 'sociologically', as a closed, in-marrying, stratum, caste is still largely confined to the Eurasian continent, or other areas where Eurasian" whites have established themselves (e.g^'Horth America, Southern Africa, t^Saharan fringes). Racial factors, which because of their visibility provide one of the most universal cards of identity used by man, also enter into the ban on intermarriage. But here
again property is heavily involved, for there appears little reluctance for men to engage in sexual unions, as distinct from marriage, with women of the lower orders. It is the sexuality of their own sisters they are concerned to protect, and the notions regarding the purity of women that attach to caste systems and the concern with their honour that marks the Mediterranean world cannot be divorced from the position of women as carriers of property. [... ]
Another kind of in-marriage is the union with cousins. Certain of these are property-conserving as far as women are concerned, especially the father's brother's daughter marriage; in societies with agnatic descent groups, this woman falls within a man's own group. [...]
Control over property can be exercised by the number as well as the kind of marriage and I predicted a connection between diverging devolution and monogamy. Where both males and females require parental'property for the maintenance^ of^Jfheir status, and where re-sources are limited^then large polygynous families are likely to have an impoverishing .effect. Only the very rich can afford_ the luxury of many children without dropping in the economic hierarchy. In dowry systems wives may be thought of as augmenting a man's wealth and hence polygyny could be advantageous to_ him; but every marriage would establish its own conjugal fund and differentiate each spouse according to the marital property she brings. There are obvious difficulties for a man in setting up a plurality of such funds (though less so when the women are sisters). The test shows a positive association between diverging devolution and monogamy. [... ]
Like monogamy, polyandry also limits the number of wives and heirs with whom the property has to be divided and this form of marriagejigain displays a positive association with diverging devolution. Indeed in Tibet the provision of one legitimate heir-producing wife for a group of brothers is explicitly thought of as a way of keeping the balance between people and land (Carrasco 1959: 36). The Atlas includes only four cases of polyandry, tliree of which are found in conjunction with diverging inheritance.
The residence of a married couple is a fuither factor likely to be influenced by property
114
JACK GOODY
INHERITANCE, PROPERTY, AND MARRIAGE IN AFRICA AND EURASIA
115
considerations; a rich wife can make a poorer husband move to her natal home; and the well-to-do father-in-law may have the same power over (and attraction for) his daughter's husband. Not every woman will display these magnetic powers, which depend essentially upon the differential distribution of wealth; it is only the rich or epicleratic daughter who finds herself in such a position. So that we should not expect a straightforward uxorilocal pattern of postmarital residence but rather an ambilocal one, where a married couple may choose to reside with either the kin of the bride or of the groom depending upon their relative position. There is another possibility: if an independent conjugal fund is established at marriage, bride and groom may also establish an independent (i.e. neolocal) residence.8
The results of this test [indicate] that, while bilocal and neolocal marriage are definitely correlated with diverging devolution, the correlation is negative where the alternative forms of marriage represent less than one-third of the total. The latter finding is contrary to my hypothesis. However the determinants of residence are not of course limited to property. Indeed in this instance it seemed possible that the absence of unilineal descent groups (i.e. 'Bilateral descent'} would prove to have a close association with the patterns of residence. [... J
I return to this question later when I compare 'bilaterality' and 'diverging devolution' as variables. Here I want to point out that it is only in respect of residence that kinship is a better predictor than the transmission of property. Part of the answer emerges from a study of the figures for the residence patterns of societies that have 'no individual property rights or no regular rules of transmission'. The interesting point is the very large proportion of these societies that include marriages of the 'bilocal' or 'virilocal with alternatives' type - no doubt because there is no immovable property to tie anyone anywhere. There are more societies with 'uxorilocal with virilocal alternative' in this category (13) than in the societies with property (9).
Property considerations, in the shape of diverging devblutionV also seemed logically connected with* kin terms. Only in exceptional cases (where there are no descendants, male, female or adopted) do brothers (or more usu-
ally brother's sons) inherit under such systems So that the three characteristics of this type o| transmission are (i) it is inter-generational (ij) it is direct and (iii) close females are preferred to distant males. Indeed devolution occurs within the nuclear family, while the establishment of separate conjugal funds differentiates sibling from sibling, parents from children1 Though cousins may be possible 'intestate' heirs, they rarely inherit. The emphasis is ori direct transmission, even if one has to adopt
It seemed possible that this 'isolation' of the nuclear family, as manifested in conjugal funds and direct inheritance, would be reflected in a kinship terminology that differentiated siblings from cousins (Murdock 1967: 158). There are several ways in which brothers (for I confine the discussion to males) may be set apart from othei kin. Firstly, under an 'Eskimo' (e.g. English) system they are distinguished from all cousins. Secondly, cousins may be distinguished not only from siblings but among themselves; this is the case with descriptive terminology ('descriptive or derivative'). [The data show a] positive correlation between terminologies that isolate siblings and diverging devolution.
While I have used inheritance, or rather devolution, as the independent variable, it is independent only in a certain context. For these hypotheses raise a further series of questions concerning the incidence of diverging devolution itself. Why should the African and Eurasian patterns be so different? T suggest that the scarcer productive resources become and the more intensively they are used, then the greater the tendency towards the retention of these resources within the basic productive and reproductive unit, which in the large majority of cases is the nuclear family. There are, several reasons for this hypothesis. Intensive agriculture, whether by plough or irrigation,.™^, permits ajmrplus of production over consurnp- \f tion that is sufficient to maintain an elaborate c? division of labour and a stratification based upon different 'styles of life'. An important means of maintaining one's style of life, and that of one's progeny, is by marriage with persons of the same or higher qualifications.9 We should therefore expect a greater emphasis upon the direct vertical transmission of prop- . erty in societies with intensive rather than extensive exploitation of agricultural resources.
This system of direct ve^tlcaljransmbs^njhe.
fro^EMSHL^2^?iy£S5'_E?n^J2.iJ:ia^:e £rov.^ sion for womenjasj^ejl as men. The position of
women in the world has to be maintained by means of property, either in dowry or inheritance - otherwise the honour of the family suffers a set-back in the eyes of others. This also means that they are likely to become residual heirs in the restricted sibling groups that monogamy permits, the property going to female descendants before collateral males, even when these are members of the same agnatic clan.
The other aspect of intensive agriculture bearing upon the conditions for the emergence of diverging devolution is the expansion of population it allows, another factor making for scarcity of land. Where intensive agriculture is dependent upon the plough, the increase in production is partly a result of the greater area a man can cultivate; once again, land becomes more valuable, especially the kind of land that can sustain permanent cultivation by means of the simpler type of plough.
Intensive exploitation of resources can be variously measured. [The data show] the firm association that exists between the presence of the plough and diverging devolution. The information on the absence of plough agriculture is not altogether satisfactory since the absence of an entry in column 39 might mean either no information or no plough; however the presence of the plough is such an obvious feature of human societies that the chance of error should be small.
The Atlas also gives a separate code for intensive agriculture (col. 28, I, /) and it is possible to retest the hypothesis on this column, though the association is somewhat less strong.
Intensive agriculture is virtually a condition of the extensive differentiation by styles of life that in turn encourages the concentration of property by inheritance and marriage. The concentration of property is maintained by diverging devolution, which takes the form of direct vertical transmission; hence the importance of 'sonship', real and fictional (which includes daughters), in these areas of social action.10 We^would therefore expect to find diverging devolution associated with complex stratified societies of all types, whether charac1-
tsri^d^bj^casxe^or^dass. This hypothesis is linked with the suggestion that diverging devo-lut^OT^jicourages^ in-marriage, which I tested earlier. Endogamy is clearly one way of limiting the consequences of the transmission of property through women. Other systems of complex stratification may restrict marriage de facto if not de jure.
The same hypothesis can be tested in column 33 which provides 'a measure of political complexity'. I have included here only the larger states. In surveying the major Eurasian civilizations, I found all (in differing degrees) to be characterized by diverging devolution: women were usually residual heiresses to their brothers, in addition to which they received a dowry if they married away. These forms of marriage prestation and inheritance are recorded in the Greek, Roman, and Hebrew and Chinese texts and in Babylonian, Hindu, and Buddhist lawbooks. For such societies were all literate; indeed testamentary inheritance, as Maine pointed out, was sometimes used to divert property from a man's agnates, who were his residual heirs. But more often and more universally the institution of adoption (often of agnates) and the 'appointed daughter' were used to ensure the direct vertical transmission of property. In general, these literate societies fall into the category of 'large states' (or are closely linked to them) and the association with diverging devolution is shown to be firm.
Alternative Predictions
To test my hypothesis, I predicted that the presence of different types of descent groups would be less closely associated with the variables examined than would the transmission of property. The runs done for diverging devolution were repeated for the nearest equivalent in kinship terms, i.e. bilateral 'descent' or kinship, that is to say those societies which possessed neither patrilineal nor matrilineal descent groups. In every case except for residence diverging devolution has a stronger association with the variables than does 'bilateral kinship'. The general point that I would make here is that while anthropologists have given most attention to kinship factors, there are a number of important areas of socfal life wfiere the mode of distributing property appears to be
116
JACK GOODY
INHERITANCE, PROPERTY, AND MARRIAGE IN AFRICA AND EURASIA
117
ire^^si^yicant. This is not a matter of trying to substitute one monolithic form of explanation for another. Nor yet of equating fas some writers have tended to do) kinship and property. The two variables are closely interlocked in pre-industrial societies where the basic means of production are almost universally transmitted between close kin. My interest is in attempting to specify the way in which they are associated, particularly with a view to ascertaining the influence of differences in systems of transmission upon kinship relations. Causal direction is not of course something one can establish by correlational analysis alone; though causal inferences can be made by means of statistical techniques it needs to be supplemented by a study of changing situations.11
I suggested therefore that 'bilateral descent' (or rather 'bilateral kinship') would be a worse predictor. On the other hand, the vesting of landed property in women, as distinct from movables alone, was thought likely to produce more significant results, anyhow as far as residence is concerned. While in every case except ^ residence bilateral kinship (i.e. the absence of unilineal descent groups} is less closely associated with the variables than diverging devolution (as I had suggested), the inheritance of land comes second to diverging devolution and is a worse rather than a better predictor. As I have earlier noted this excludes societies with a landed dowry (and in inheritance of land by women) which may well be a better predictor; but I make no attempt to explain this result.
Conclusions
Tylor long ago pointed out the adaptive functions of exogamy for human societies (1889). Mankind, he remarked, was faced with the alternative of marrying out or being killed out. In-marriage on the other hand is a policy of isolation. One reason (and there are of course others) for such a policy is to preserve property where this is transmitted through f-both males and females, to encourage marriages with families 'of one's own kind' and thus to maintain property and prestige. The positive control of marriage arrangements (ex- ' ogamy is a negative control) is. stricter where 0 property is transmitted to women. It is a commentary on their lot that where they are more
propertied they are initially less free as far ls marital arrangements go, though the unions into which they enter are more likely to 1 monogamous (or even polyandrous).
In this paper I have_tried to test by means of the JMrnographic Atlasja set of hypotheses to do with the concomitants of diverging devolution derived from a more intensive study of the liEl erature on a number of societies in Eurasia and Africa. Though the information is imperfect and the instrument limited, my basic suggestions are all confirmed (except for caste). Bilateral kinship was a worse predictor except for certain types of residence; on the other hand the inheritance of land was also worse, which' the theory did not anticipate. These results are reinforced by the fact that the hypotheses are interlocked and were formulated in advance.
I have sjrown that^Em so„ cieties differ in their systems p£transmitting property; ^these differences are i^orxejated with differences in the types of marriage prestation, the extent of control over \vornra,_both before nwriage_and in terms qf_ marriage^r^rtngrs (and probably, too, after marriage though this I could not test). Differences in the nature of a man's estate are indicated in the greater prevalence of terms that isolate the sibling group, an indication of the differences in the type of descent corporation found associated with the systems for the devolution of property and with modes of agriculture. Both the means of production and the relationships of production have certain marked differences.
It is a failure to recognize these differences in the type of descent corporation (even when both can be described as 'patrilineal descent groups') that seems to have caused much of the controversy over the application of descent or alliance models to the study of such societies, and it is significant that so-called 'descent theorists' have generally worked in societies with homogeneous devolution and the 'alliance theorists' in societies with diverging devolution. It should be apparent that where marriage involves a re-arrangement of property rights of the dowry kind, then conjugal, affinal^rsjbling, and filial relationships. are_ likely_tq,„display qualitative differences from systems of the African ..kind.. It also follows that the organization of descent groups will differ under these two conditions. The ball-play between rival 'theor-
ists' has obscured the basic differences in the material they are dealing with.
I shall pursue this point in another place; meanwhile I conclude by pointing to the association of diverging transmission with intensive (and plough) agriculture, with large states and with complex systems of stratification. In such societies social differentiation, based on productive property, exists even at the village level; to maintain the position of the family, a man endows (and controls) his daughters as well as sons, and these ends are promoted by the tendency towards monogamous marriage.12 Indeed it is significant that the strongest associations of diverging devolution are with monogamy and plough agriculture.
NOTES
Note: The data presented in this paper have subsequently been analysed by a type of multivariate analysis, in fact a combination of pathway and linkage techniques. The results are embodied in a further paper, Causal inferences concerning diverging devolution, by Jack Goody, Barrie Irving, and Nicky Tahany, which it is hoped to publish shortly.
1 Bosman was accustomed to the conjugal community of Roman Dutch Law. But the absence of a dowry was apparent even to Englishmen reared under the qualified unity of conjugal property which was a feature of their Common Law. This broad distinction between Africa^nd^Jau^pejtlil^ 'MfF"
j    riage in Ashanti between free persons.^.does J    not jead to community of property, between f    the spouses, still less to the_sjnHng^^_the wife's legal persona in her husband's' (Allott 1966:191). '
2 The absence of a dowry is enshrined in local usage of English and French. For the Europeans, having no word for prestations that passed from the family of the groom to that of the bride, used the word they had, 'dot' or 'dowry', for a very different set of transactions. These terms are still sometimes used for 'bridewealth' and other prestations.
3 Barton used the term 'homoparental' but this presents difficulties for matrilineal inheritance.
4 I have used all the data from the Atlas and have not sampled them in the way suggested
by Murdock, 1967:114. The reasons for this are purely practical and I shall conduct further tests when time allows.
5 The phi coefficient is also known as the Kendall partial rank coefficient (Siegel 1956: 225-26).
6 In speaking of the 'woman's property complex' I refer to her access to property held by males. Under systems of homogeneous inheritance, women have property but it is either inherited from women orself-acquired.
7 Property was also destroyed during the lifetime of the holder in the well-known case of the potlatch.
8 Logically, neolocal residence is less firmly attached to diverging devolution than are bilocal or the less evenly distributed systems of alternative residence. But the boundaries of 'neolocal' residence are difficult to discriminate (how separate is 'separate'?).
9 In my account of the LoWiili of Northern Ghana (1956), I suggested that the increased differentiation of wealth militated against the movement of property outside the co-residential group. The specific feature of this group that inhibits a dispersal of the property is its character as a unit of production. Where self-acquired property (income) begins to play a greater part than inherited wealth (capital), then there will be increasing reluctance to allow property to go outside, as is bound to be the case where the residential pattern is incongruent (or partially so, as in a fully fledged system of double descent) with the mode of inheritance.
10 This point is developed elsewhere (Goody 1968a).
11 I have tried to approach this problem in my comparative work on the LoWiili and the LoDagaha, as well as in a separate essay (Goody 1968b).
12 This paper represents the preliminary results of the comparative research I have undertaken with the help of a grant from the Social Science Research Council. I have hoped for some years to carry out such a survey but it would have been impossible but for the help, by means of this grant, of my assistant in research, Mrs. L. March. I should also like to thank Dr. L. Slater for the programming, and Dr. E. N. Goody, for comments and advice. Little systematic comparison has been carried out in Europe since the work of Hobhouse, Wheeler, and
118
JACK GOODY
1
Ginsberg (1915); like all recent writers on the subject I am much indebted to G. P. Mur-dock for making his collection of data available in the shape of the Ethnographic Atlas.
BIBLIOGRAPHY
Allott, Anthony N. 1966. The Ashanti law of property. Zeitschrift für vergleichende Rechtswissenschaft. 68:129-215.
Bosnian, William. 1967. A new and accurate description of the coast of Guinea. London: Frank Cass (First English ed. 1705).
Carrasco, Pedro. 1959. Land and policy in Tibet. (Am. Eth. Soc.}. Seattle: Univ. of Washington Press.
Driver, Harold Edso. 1961. Introduction to statistics for comparative research, Readings in cross-cultural methodology (ed. Frank W. Moore). New Haven: H.R.A.F.
Goode, William J. 1963. World revolution and family pattern. New York: Free Press.
Goody, Jack. 1956. The social organisation of the LoWiili. London: H.M.S.O. (Second ed. 1967}.
Goody, Jack. 1958. The fission of domestic groups among the LoDagaba, The developmental cycle in domestic groups (Cambridge Papers in Social Anthropology, No. 1). Cambridge: University Press.
Goody, jack. 1959. The mother's brother and the sister's son in West Africa. /. R. Anthrop. Inst. 89: 61-88.
Goody, Jack. 1962. Death, property and the ancestors. Stanford: University Press.
Goody, Jack. 1968a. Adoption in cross-cultural perspective. Comparative Studies in Society and History.
Goody, Jack. 1968b. Inheritance, social change and the boundary problem, Levi-Strauss Festschrift (ed. J. Pouillon and P. Maranda). The Hague: Mouton.
Goody, Jack. 1969. Marriage policy and incorporation in northern Ghana, From tribe to nation in Africa (ed. R. Cohen and J. Middleton). San Francisco: Chandler.
Hobhouse, L. X, Wheeler, G. C. and Ginsberg, M. 1915. The Material Culture and Social Institutions of the Simpler Peoples. London.
Murdock, George P. 1960. Cognatic forms of social organization, Social structure in southeast Asia (Viking Fund Pubs, in Anthrop. No. 29). Chicago: Quadrangle Books.
Murdock, George P. 1967. Ethnographic atlas: a summary. Ethnography 6: 109-236.
Redfield, Robert. 1934. Chan Kom: a Maya village. Washington D.C.: Carnegie Institute of Washington.
Siegel, Sidney. 1956. Nonpammetric statistics for the behavioral sciences. New York: McGraw-Hill.
Tylor, Edward B. 1889. On a method of investigating the development of institutions: applied to laws of marriage and descent. /. Anthrop. Inst. 18: 245-269.
Section 2  Terminology and Affinal Alliance