Implications of Endogamy in the Southwest Eurasian Highlands: Another look at Jack Goody’s theory of production, property and kinship

ABSTRACT Reviving interest in Jack Goody’s comparative and historical work on systems of production, kinship and state organization prompts us to revisit his analysis in ‘Production and Reproduction’ – looking both at the data and at the explanatory framework. While we accept Goody’s emphasis on the central importance of productive technology, we nevertheless contend that he underplays the role of kinship and marriage systems in fostering agricultural production, the growth of state power and inequality. We argue that this oversight results from Goody’s ‘resolute materialism’ and that a more realistic view requires a more systematic formulation of the interactions between pragmatic rationality and the logic of social identity and cohesion. We develop an alternative argument, adding a Maussian analysis of exchange and identity and evaluate its implications using Goody’s data and our ethnographies. Finally, we consider the implications this raises for future research – and how these might be answered by ethnography and comparison.


Introduction
In a review of Goody's intellectual legacy written more than ten years ago, Hann (2008) focused on Goody's comparative work on inheritance systems, and their connections to many aspects of productive and social lifepraising its intellectual scope, geographical and temporal rangeand the willingness to engage with statistical methods and theories of social evolution. Hann also noted however that Goody's work was theoretically and methodologically unfashionable among social anthropologistsbut hoped, given the importance of its themes, and their relevance to present-day political developments, that this situation might be about to change.
Since the time Hann wrote, the themes that Goody discussed and Goody's own contribution, have indeed received renewed attention. The revival has come partly from archaeologists and has owed a great deal to technical developmentsincluding new ways of assessing biological relatedness and place of origin from human remains (Ensor, Irish, and Keegan 2017), and inferring kinship structure and degrees of inequality from the relative sizes and arrangement of buildings (Ensor 2017;Kohler et al. 2017). The renewed interest has also come from evolutionary anthropologists, who have pioneered 'phylogenetic' ways of projecting social arrangements back into the past, based on an analogy between linguistic change and the Darwinian development of separate but related species. Working with data on Indo-European speaking societies, Fortunato and colleagues have used this approach to revisit several of the themes tackled by Goody (Fortunato 2017;Fortunato, Holden, and Mace 2006).
However, so far this revival has had little impact on social anthropology itself. Writing as social anthropologists who sympathize with Goody's agenda, we want to contribute to the reviving comparative and meta-historical agenda from the perspective of social anthropology. In order to re-examine Goody's agenda, we refer to another comparative thinker and re-work and extend Goody's model through Maussian theory of gift exchange. Goody himself argued that comparative methods and ethnographic research should be complementary rather than alternative approachesand neither is likely to yield its full potential without referring to the other. As we will show below, this implied warning applies as much to Goody's own comparative work as to anyone else's. An ethnographically inspired review of his analysis raises issues that deserve further explorationboth by further ethnographic research and by using the developing techniques of comparative and archaeological research.

A critical discussion of Goody's thesis in Production and Reproduction
The most important single publication in Goody's comparative opus is Production and Reproduction (1976, henceforth PAR). The work starts by noting a contrast between African and European (more generally Eurasian 1 ) arrangements concerning family property and inheritance. In Africa husbands and wives typically have separate property, and on the death of either partner, the deceased person's property passes to members of their family (or rather lineage) of origin. Property therefore always remains within the same lineage. In Europe husbands and wives typically hold their property jointly, and some of this property is inherited by both sons and daughtersdaughters often receiving some of this as a dowry when they marry. If one is thinking in terms of lineages, (more particularly of patrilineages), property inherited by daughters passes out of the lineage groupand is therefore referred to by Goody as diverging devolution (Goody 1976, 6-8).
Goody was interested in the contrast between the African system of lineal inheritance and the European (more generally Eurasian) system of diverging devolutionand wants to understand their causes and consequences. He was particularly interested in their consequences for kinship practiceswhat he refers to as the domestic domainincluding such variables as monogamy versus polygamy, restrictions on pre-marital sexual behaviour, and the choice of marriage partner. This is the issue on which we will focus here: in particular the contrast between the rule of lineage exogamy, which is virtually universal in sub-Saharan Africa, and the practices of marrying within the same groupwhether by social class, caste, locality or lineagewhich Goody refers to variously as endogamy, homogamy and in-marriageand which is commonly, but not universally, found in Eurasia.
There are two points to note about the way the table is set out. The first is that causation runs from left to right: from Economy as the first cause, to Marriage System as the final effect. The second point is that causation proceeds one column at a timeso that any causal relationship between variables in non-adjacent columns is assumed to be due to their mutual relationship to the variable or variables in the column in between. The fact that diverging devolution is the only variable in column 3 expresses Goody's belief that it provides the sole connection between in-marriage and the other variables in the model. Goody argues that this model should be seen as a first approximation both to the underlying causal relationships, and to the likely (pre)historical order in which the developments first occurred (Goody 1976, 8, 37-40).
To grasp the essence of Goody's explanation, it is best to start with the proximate cause of in-marriage (i.e. with the transition between columns 3 and 4 at the right-hand end of the table). Goody argues that there will be a strong association between diverging devolution and in-marriage. The reason for this is that, given the existence of diverging devolution, parents need to control their daughters' marriages, in order to prevent them transmitting property outside the social group.
But why are daughters endowed with property in the first place? Goody offers two explanations (depicted in the possible transitions between columns 2 and 3) both of which could account for the widespread presence of diverging devolution in Europe, and its virtual absence in Africa. The first is that, in a property-stratified society, it is necessary to endow daughters in order to enable them to make an equal match. The second argument (which derives from the work of Boserup (1970)) is that if the husband contributes more labour to the family economy, the family of the wife needs to balance the husband's labour by contributing more property.
The ultimate cause, however, is the development of advanced agriculturein Eurasia, 3 but not in Africawhich (as depicted in the transitions between column 1 and column 2) demands a high input of male labour, and leads to the development of property stratified state societies.

Outline of the statistical evidence
The evidence comes from Murdock's Ethnographic Atlas (Murdock 1967) which is a database, compiled from ethnographies and historical accounts, of codes representing selected characteristics of 863 4 societies. The relevant findings in terms of correlations between different factors are summarized in Table 2.
We would like to make a general point about this correlation matrix: all the coefficients are positive. This means that the variables which, according to Goody's theory ought to go together, do in fact go togetherwhich shows that the theory does at least have a certain basic credibility. It is important to bear this overall finding in mind, because in the next few paragraphs, we will present some serious criticisms of particular aspects of the theory. But this overall result says nothing about the specific causal connections between the different variables, or the overall causal sequence. At this point any investigation based on synchronic correlations faces a well-known difficulty. This is that, although correlations can strongly suggest the existence and strength of specific causal connections, they cannot determine their direction. Nevertheless, they do provide important clues about the potential validity of the model. In order to extract these clues what Goody did (or rather what J.C. Mitchell did on Goody's behalf (Goody 1976, 37)) was to carry out a path analysis that used the correlation matrix to estimate the strength of the connections between the different variables, on the assumption that the overall causal order in Table 1 was correct. By comparing the strength of the observed correlations with those predicted by the model, it is possible to assess which aspects of the model have most empirical support, and also to identify places whereregardless of the direction of causationadditional hypotheses would need to be included in the model in order to explain the empirical findings. The results, which were set out in diagrammatic form on page 39 of PAR, did not entirely confirm Goody's expectations. The first major difference is that diverging devolution is much more strongly connected with male farming than with complex politysuggesting that Boserup's (1970) explanation is nearer the truth than Goody's original status-conservation theory. The second is that, although diverging devolution does indeed provide a causal link between in-marriage and the two variables in column 2 of Table 1 (complex polity and male farming), there are also direct causal connections between in-marriage and the two other variables, which are independent of diverging devolution. In other words, Goody's own data suggest that, although there does appear to be a causal link between inheritance systems and endogamy, endogamy is also connected with farming systems and with political organization in other ways which are not due to property strategies. 5 Ethnographic critique of Goody's model These quantitative findings are consistent with two lines of criticism that emerge from the ethnographic and anthropological literature. The first concerns a mismatch between Goody's emphasis on the processes of agricultural intensification, state formation and social stratification. These processes affect all the advanced societies of Eurasia, and the  this table are reproduced from Table 22, on page 132 of Production and Reproduction (Goody 1976). b This is a type of correlation coefficient that is appropriate for binary variables.
need to preserve status is, by definition, strongest among the rich and powerful. The ethnographic evidence however is clear that endogamy is particularly marked in the circum-Mediterranean area (Bonte 1994)and also that it is often not associated with social stratification. Several studies of endogamous circum-Mediterranean communities stress that it is accompanied by an ethos of equality and that local endogamy is, if anything, more characteristic of less prosperous families than of the richer members of the community (Barth 1953;Bates 1974;Khuri 1970;Leach 1940).
The other line of criticism concerns what Hann (2016) refers to as Goody's 'strong materialism'. As we have seen, in Goody's causal model, endogamy is very closely tied to considerations of property. But it can also be viewed in other ways. Some authors (e.g. Pina-Cabral 1992) see endogamy as primarily an expression of social unity. It can also be seen as a conceptual system: a particular way of understanding the nature of marriage and the relationships to which it gives rise. Indeed Goody himself treats it this way in the opening pages of PAR, though that part of his discussion is not reflected in his causal model.
Despite slanging matches about 'vulgar materialism' and 'vulgar idealism' (Hann 2017, 235), these different interpretations of endogamy need not be incompatible. They can be inter-related in various ways, and all can potentially be linked to other aspects of social and economic organization in order to account for the presence or absence of endogamy in different societies. However it is clear, from the statistical evidence discussed above, that the causal model's exclusive concentration on the causes and consequences of diverging devolution does not provide a complete explanation of the connections between endogamy, agriculture and socio-political organization.
In the next section of this paper we present ethnographic data from two societies in which female inheritance does not play an important role, in order to explore what other connections there might be.

A natural experiment: Endogamy without diverging devolution
South-east Anatolia and north-east Italy -Similarities between two highland societies The data comes from our own first field sites: the Kurdish village of Sisin in the southeastern Turkish province of Hakkari (Yalçın-Heckmann 1991) and the Romance speaking village of Ovasta, in the sub-region of Carnia in north-east Italy (Heady 1999). Both villages had populations of about 175-250 in the 1980s. They were situated in mountainous areas, with traditional economies based on a mixture of farming and transhumant herding, and in each case an important role was played by common property held by the village as a whole. In both places, post-marital residence was strictly patrilocal, and property was inherited almost exclusively by males, with brothers receiving equal shares. Though the ideologies of marriage differed, they both attached a positive value to locally endogamous matches, without making them an exclusive rule. In both places, unions between a man and a woman from the same village amounted to about half of the resident married couples (Heady 1999, 5-8, 25-35, 131-134;Yalçın-Heckmann 1991, 228-241).
Since several of these characteristics are shared with other circum-Mediterranean societies, these two field sites provide a kind of natural experiment which may also throw light on the reasons for the high levels of endogamy in this part of the world.
As a first step, we present the data on endogamy itself. Tribal endogamy was 100% within Sisin. 6 Among all 65 existing marriages in Sisin at the time of fieldwork, 22 of them involved couples who were not cognate/agnate to one another, though they were nevertheless members of the same tribe. All other marriages were between kin. The figures for Ovasta are as follows: at the time of fieldwork sixteen of the 44 existing marriages (36 percent) were locally endogamous. Nearly all the rest were patrilocal, with non-Ovastan women moving to Ovasta to join their husbands. The endogamy rate had been declining in recent decadesfrom 10 (53%) of the 19 marriages contracted before 1960, to just 1 of the 8 marriages contracted between 1980 and 1991and the changing marriage patterns were associated with other social and economic changes. In the discussion that follows, we will focus on the system as it was when endogamy was still frequent.
We would now like to describe how the two systems workfocussing particularly on their implications for status, conflict and cooperation. The discussion is structured primarily around the axes of (a) seniority, (b) gender, (c) economy, and (d) military activities. The first two axes involve linked principles of hierarchical classification and are related to the system of social meanings associated with endogamy. The other two axes involve institutional spheres in which individual men, and their families, compete for personal status, but in a way that accepts and reproduces an underlying principle of categorical equality and cooperation.

Status, unity and the gendered life-cycle
We begin with seniority. In Hakkari, as in many Muslim and patrilineal societies, seniority meant higher status (Yalçın-Heckmann 1991, 149-167). Senior men were respected in the society, they were given better and first food, better places to sit, were addressed with respectful terms, and in general were treated differentially and had an advantage over younger men and certainly over women. There were, however, two aspects of this seniority: it was dependent on the life cycle, hence all men could expect to be respected with advancing age. The second aspect is that respect was also associated with the lineage and clan structure and historical significance, taking the form of lineage or clan honour, as has often been discussed in relation to clan 'sharaf' (Barth 1953;Eickelman 1998;King and Stone 2010;Leach 1940;Meeker 1976;Peters 1967). This kind of honour and respect was due to all men (and women) who belonged to a particular tribal category, hence it worked as a levelling mechanism among men (but not as much between men and women), among those who were of course members of this particular lineage/clan.
In the village of Sisin, in Hakkari, solidarity among men was framed within the conceptual model of patriliny. 7 The Kurdish kinship terminology was used within the tribal system to express this sense of unity by referring to all patrilineal members of the lineage and clan as classificatory patrilineal cousins (pismam) 8thereby creating not only a feeling of intimacy and closeness (Cuisenier 1975;Eickelman 1998) but also, we argue, structural equality, as has been well documented in segmentary societies (Evans-Pritchard 1970;Peters 1967;Tillion 1983Tillion [1966). The sense of patrilineal solidarity could be stressed even more stronglyby making the rhetorical claim that 'we are all one man'especially when one wanted to highlight the need to prevent dissent within various levels of tribal segmentation.
This patrilineal model, was, as King and Stone (2010) would expect, closely linked with the performance of masculinity (mêranî in Kurdish, see also Herzfeld (1985) on similar performances of mountain Cretan manhood, or Gilsenan (1996) on Lebanese rhetoric of manhood), and with strong notions of patrilineal continuity (xwîn in Kurdish, soy in Turkish). It was also linked to the subordination of women. In Hakkari, women's fertility, labour, marriage choices, and inheritance were strictly controlled by men. Women in Hakkari's semi-nomadic tribal society were born into and grew up in large households, where female and male chores were strongly differentiated and women worked very hard and their work was essential to household's economic livelihood and marriage strategies. Although gender segregation as known in some Muslim societies was not very significant (cf. Tillion 1983Tillion [1966), patriarchal and discriminatory practices against women such as physical violence, honour killings, and early/child marriage practices were not uncommon.
Yet, women in these tribal contexts were also valued and significant as individuals. There were even cases where women became the name givers to lineages. And women were also valued for making links between otherwise patrilineally and descent divided lineages. 9 Quite apart from these prominent examples, the reason given for endogamous marriagethat they didn't want to give their girls away to strangersexpresses a sense that young women are valuable, and all the more so because they are already members of the group.
They were however subordinate, as indeedif rather less sowere their brothers. Young men had few possibilities of challenging the senior men who were in control of property and women. One had access to these 'valuables' as one went through the life cycle of growing up, going to the military, marrying, having children, and finally separating one's own household either from the parental or the fraternal household.
There was however one way in which young men could challenge their subordination, particularly to the marriage plans made for them by senior men (cf. Bates 1974). Bride-kidnapping was a common strategy, of which the local interpretation would vary between elopement with consent or abduction without the consent of the woman. This was a situation which put the symbolism of patrilineal unity and male honour to one of its strongest tests. Especially in cases where a woman was thought to be agreeing to the kidnapping man's will (hence elopement, in anthropological jargon), the male relatives of the woman would acknowledge the challenge by taking up arms and chasing the kidnappers, but they would shoot in the air as in mock fighting between tribally close segments. If the wife-kidnappers were from segments of the tribe, which were structurally close to the lineage of the kidnapped woman, or if the kidnappers were from affinally related lineages and households or were neighbours (all cases where close group solidarity would be the essential basis for collective action), the mediators of the conflict would use the following formula most often: em mêrovê tekin, 'we are all one man'.
Much of this discussion carries over unchanged to Carnia. Here too, membership of the community was based on descent through maleswith the same implication of strictly patrilocal marriage. Households were multi-generational and extended, with a clear gendered division of labour, and authority vested in the older generationwith the senior woman having authority over daughters and daughters-in-law, and the senior man having authority over his sons and the household as a whole. In Carnia, as in Hakkari, there was a strong emphasis on the importance of local unityand villagers all shared in the collective status brought by the visible prosperity and reputed solidarity of their village. As in Hakkari, the community was linked together by ties of both descent and marriageand both kinds of tie have a role in the symbolism of village unity. However, a different emphasis is given to each kind of tie.
To start, let us look at descent. While in Hakkari the descent tie is treated as the primary source of unity, in Carnia it is not. Most people do not trace their descent back more than about four generations, and the inhabitants of any particular village do not see themselves, even ideally, as the patrilineal descendants of a single male ancestor. Instead, the village is seen as united by ritual kinshipties of individual and collective god-parenthood (Heady 2018). This is still descent of a kind, but it does not embody the patrilineal principle that underlies collective identity in Hakkari (and family identity in both places).
If the emphasis on descent is weaker in Carnia, the stress on unity through marriage is stronger. Carnia shares with Hakkari the reluctance to give girls away to strangers, and also the sense that the exchange of brides can create ties of solidarity. But instead of a restricted exchange of brides taking place between two families, the Carnian system involves the generalized exchange of brides between all families of the village. 10 This is linked to another difference: the choice of marriage partners was a matter for the young people of the villagenot for their families. As we have seen, this possibility exists in the form of kidnapping/elopement in Hakkari: but while it is an alternative form in Hakkari, it was the norm in Carnia, where parentally arranged marriages were rare exceptions.
To make this possible, both the practical and symbolic arrangements for socializing between young men and women were very different from those in Hakkari. Between their late teens and marriage, young people of both sexes formed a recognized social group within the village expected to spend time together free from the supervision of the older generation. Honour killings were unheard of, but young women did face public shaming for sexual immoralityin two contexts. The first was if a girl cheated on her regular boyfriend, in which case the shaming was done by the young men of the village. The second was if she became pregnant and could not persuade the child's father to marry her. The role of young men as controllers of courtship was expressed in an annual ritual in which young unmarried men entered every house in the village, and thenwith the assistance of the young womenorganized a dance for the village as a whole. After the house visitbut before the dancethe young men publically announced the names of courting couples, and the parents were obliged to allow their daughters to be escorted to the dance by the partner who had been announced by the young men (Heady 2003).
A final angle on the social meaning of marriage is provided by the accompanying payments. Women in Hakkari did not bring in into marriage any property as trousseaus, but as long as there was endogamy there was no bride-price paid either. 11 More significantly, the lack of trousseau and bride-price was explained as maintaining equality between the tribal members and as not giving daughters away to strangers. The principle of equality among tribesmen was amplified in the marriage practice of direct exchange of marriage partners, where two women moved between two different households as wives of two men in these households.
In Carnia too, no bride-price was involved for marriages within the village community, but a symbolic bride-price was demanded for marriages to outsiders. However, the brideprice was paid, not to the girl's family, but to the young men of the village (who spent it on a big party). Brides did not bring dowries in the form of housing or land into the marriage, but they did bring trousseauxin the form of clothes, linen, work tools, and also the mattress for the marital bed (which had been jointly sewed by the bride and her women friends). Implicitly, these goods would help to form the basis of a new conjugal family, which would not be entirely under the control of the senior generation.

Economy
It can be seen from the above illustrations that economy, our third axis, is closely linked with the marriage system in both cases. Both in Carnia and in Hakkari, the management of commons was corollary to solidarity as a village unit (Carnia) or as a tribal unit (Hakkari). In the latter case, tribal pastures were the central common property, for which ownership was customarily defined as tribal lands (and legally held by the state), but use rights were collectively managed and strongly protected. In fact, the collective protection of 'traditional' tribal pastures (which were central to the herding economy at the time) made the backbone of and the only reason for collective tribal action. In no other circumstances would the whole tribe come together.
At the time of fieldwork in Sisin the economy was strongly relying on semi-nomadic sheep and goat herding, supplemented with intensive agriculture of terraced fields in the villages. The pastures and hay cutting areas were claimed and managed by the tribe. The usage rights were distributed along the sublineages and managed collectively by households belonging to these sublineages. The terraced fields and land for building houses, however, were given by the government in the 1940s and became individual property to be inherited among male descendants. With population increase and limited availability of cultivable land, the need for cooperation between tribal villages was met partly by keeping the rule of endogamy within the tribe and exchanging wives between tribal villages. Even if women were not equals and hence did not have use or inheritance rights to pastures, affinal ties through women often played a role in securing peace and/or access to pasture and camping grounds.
At the time of fieldwork in Carnia the village commons, and the economic system of which it had been part, no longer had much practical significance. The tipping point in this process of economic change occurred in the early 1970s. But the earlier system, and the role within it of common village property, had been living realities until that timeand so it was possible to reconstruct how the system had worked by speaking with middle-aged and older informants.
The economic situation in Carnia before the 1970s was not unlike that in Hakkari, though several differences are worth noting. In Carnia, forestry was as much of an economic resource as cattle raising, and until the eighteenth century both the forests and the high pastures had been the collective property of individual villages (Bianco 1985). Ovasta was one of a minority of local villages in which this was still the casethough other villages as well maintained some informal use rights in their former commons. Just as important as common lands was a tradition of collective enterprise. Almost every village had its own cooperative dairy for processing the milk into cheese and selling on some of the product to outside traders. Many also had commonly run stalls in which the village cattle were kept during the winter. Gathering firewood and maintaining paths to the high pastures were also managed as collective village tasksas was keeping the roads free of snow during the winter.
Nevertheless, despite the importance of collective resources and activities, each family was also its own business unitboth in Carnia and Hakkari. In Carnia this involved harvesting hay and vegetables from its own land, keeping its own animals, and also drawing a good deal of income from some family members' participation in the monetary economywhether as local forestry workers or sawmill operators, or as seasonal emigrants.
In Hakkari successful household management involved managing the herd size, controlling male and female labour within the household, and controlling marriage strategies. The economy of mountain villages in Hakkari had limited use of money, which was acquired from the sale of butter and sheep wool, also of sheep when there were major expenses. Otherwise, subsistence economy dominated in the early 1980s. All agricultural produce was either used by the household and/or bartered with wheat and vegetables grown in plains villages. Goods and products which were not produced by households (consumption goods as well as household and construction goods, gadgets, and simple machines) were bought by money, either from the local markets or smuggled across the Turkish Iraqi border. Despite all these economic relationships with townsmen and agriculturalist villagers, marrying into town or non-tribal agriculturalists or taking wives from them was very rare. 12 In both places village families were trying to maintain a reasonable living standard (given the expectations of those days) but their economic efforts also had social meanings. Economic achievement was also important for the family's social position in the village as a whole. In Carnia it was (and is) important to have a house that looked good, but also to keep one's fields neat and tidy and to keep up with the neighbours in starting and completing each season's agricultural tasks. In both places, men strove to accumulate social and economic capital through strategical successful household management. The possession of rights in a home, and of the land needed to support a family, was crucial for male statusand for this reason houses were transmitted exclusively to sons.
But this competition, though intense, was not unlimited. Economic cooperation in cultivating and herding were important reasons to stick together instead of distinguishing oneself from others (for Kurds cf. Leach 1940, 21). People took turns working together on each other's landand it was also important to help out people in particular need without calculating an immediate return. Looking back, people in Carnia saw this as a time of both competition and solidarity in which no-oneexcept for a few rich families (one or two in each village)was much better than anyone else. This attitude affected the way that middle-aged and older people around 1990 viewed the behaviour of younger people who, because of the rapid decline of agriculture during the previous two decades, no longer helped out much in the fields. The older people thought their juniors considered themselves superior -'like rich people'no longer willing to join in the egalitarian work relationships that had been central to their parents' lives (cf. Bailey 1971 'competing to remain equal'). Indeed, cooperation was central to the continuance of the village as a social and reproductive unit. In Carnia, once the collective pressure to cooperate on commons diminished, people also stopped marrying within the same villageconfirming that endogamy, cooperation, and the sense of equality were complementary aspects of a single social system.

Military activities
Our final axis concerns military life. In both Carnia and Hakkari, the practice of conscription had been incorporated into local folklore as both a masculine rite of passage and an affirmation of the on-going vigour of the community as a whole. In Carnia, many family homes displayed photos of their sons or fathers in the uniforms of the mountain infantry (Alpini) or mountain artillery: regiments whose traditions and uniforms celebrated their own mountain identity. Conscription was also incorporated into the symbolism of the annual ceremony in which the young men asserted their ritual control of courtship. They became eligible for membership of the young men's group once they had reached the age of military service and attained the status of 'coscrits' (conscripts).
In Hakkari, military conscription meant serving in the Turkish army, which was seen as a rite de passage and was also a process of experiencing one's own male and Kurdish identity outside the tribal area. Similar to Carnia, young conscripts (who were called to service as cohorts) left together after visiting all the village households and saying farewell and receiving money gifts; on their return they again visited all the village households and were welcomed with money gifts, resonating very much van Gennep's description of separation and integration in rites of passage. Young men, therefore, were first separated from the village and tribal community of equals and were ritually re-integrated as adult men upon their arrival. Their military experiencelike in Carniawas later commemorated with their formal pictures in uniform hung on each household's home.
It might seem, therefore, that the local community was dependent on the state in this crucial symbolic respectand there is clearly some truth in this idea. However, there is another implication that is equally important: that the local community is itself a society of soldiers, capable, if necessary, of independent military action. This is plainly the case in Hakkari, whose rather lawless recent history has often involved local communities in military action (cf. Leach 1940, 55). In Hakkari, the prime military unit is the multi-village tribe which, unlike the village, is more strictly endogamous. Indeed, the need to maintain marriage links between different villages within the same tribal area may explain why individual villages are only partly endogamous.
Carnia has no social unit corresponding to the tribeand the greater degree of state control than in south-east Anatolia has usually meant that there was less scope for purely local military action. Nevertheless, there is a sense of shared Carnian unity, and whenin the closing phase of the Second World Warthe opportunity arose for local military action, the Carnians took it: declaring their independence from the German and Fascist authorities, and maintaining this independence for several months (Angeli and Candotti 1971).

Extending Goody's argument
Our underlying aim in reporting this 'natural experiment' was to identify causal links between endogamy and other socio-economic factors, which might have been missed by the model set out in Table 1. We were looking in the first instance for factors which could explain the presence of endogamy in societies without diverging devolution. Our hope was that the factors we identified would not be restricted to such societies, and would improve the general explanatory and predictive power of Goody's model in other societies as well. So in the discussion that follows we will consider both local effects and the possibility that the effects might be relevant more widely.
The first step, which is the theme of this section, is to identify the factors concerned, and review the range of their potential effects. The second step, which will be tackled in the next full section, is to formulate these effects as a theoretical model. The third step is to apply this model to Goody's historical account, and outline how the revised version can be checked against statistical data in the same way as Goody's own analysis in PAR.

The functional importance of endogamous kinship ties
Perhaps the most striking finding to emerge from our natural experiment is the practical and symbolic importance of the sense of shared kinship in both Hakkari and Carnianot just for individual behaviour and family organization but also as the basis for collective action within and between local communities. The actions concerned include . peace-keeping within the local community . organizing access to and maintenance of common lands and other collective resources . local defence.
These forms of joint action provide an environment in which the main way of achieving status is peaceful agricultural work and the accumulation of wealth.
Endogamy is closely related to the sense of mutual relatedness. Indeed, in Carnia local marriage ties are celebrated as the main basis of shared kinship. Although this is not the case in Hakkari, where local unity is understood primarily in terms of shared patrilineal descent, the point that stands out in both places is that endogamy is seen as a natural consequence and confirmation of the value attached to local kinship ties.
These findings may help to fill the gaps in Goody's statistical model. The main statistical problem that we noted earlier with the model in Table 1, was that 'In-Marriage' was more strongly correlated with the variables 'Male Farming' and 'Complex Polity' than could be explained by the indirect connection through 'Diverging Devolution' which was all that was allowed for in Goody's model. The finding that endogamy limits conflict and enhances social cohesion and belonging, may help to explain its empirical association with political development. Combined with the resulting tendency for men to focus their status-seeking efforts on economic work, these results might enable us to add direct connections between 'In-Marriage' and 'Male Farming' to the causal model in Table 1.

Function, causation and strategy
But establishing the functional importance of the sense of shared kinship is not the same as explaining how it arisesor at least not entirely. Function and cause may be closely linkedif the reason why a particular social arrangement or set of symbols is adopted by the members of a community is that they believe it will bring them practical benefits. The adoption of the symbolism, or the fact, of in-marriage could then be explained as a deliberate strategy. This interpretation is supported by the changes in Carnian social life since the 1960sthough in this case the strategic nature of collective kinship practices was indicated by the timing of their abandonment. As we noted above, the decline in the practical importance of local agriculture, and the practical cooperation it involved, was accompanied by a sharp fall in village endogamy (which was followed by some weakening of the symbolism of local relatedness). The differences between the kinship systems of Hakkari and Carnia may also be explained as the outcome of strategic choices. The greater emphasis on patrilineal organization in Hakkari is consistent with the semi-nomadic economy, in which the need to protect the family herds may have favoured close cooperative ties between related males. But strategic explanations, though entirely valid, can only take us so far. They explain when kinship solidarity is likely to be invoked, but not why shared kinship is associated with solidarity in the first place.

Theorising the power of kinship: Axioms of identity and amity
This question is, of course, fundamental to any kind of kinship theoryand it is remarkable how similar are the answers proposed by different theoretical schools. Cultural anthropologists (Sahlins 2013;Schneider 1980), evolutionary biologists (Hamilton 2006;Hughes 1988), and social anthropologists (Fortes 2006(Fortes [1969) all agree that kinship is a kind of shared identity which carries with it a presumption (other things being equal) of mutual solidarity. Fortes's formulation of his 'axiom of kinship amity' is particularly helpful as he makes it clear that the identity concerned is not just a matter of genealogical connectedness but of the ways in which genealogical connections are classified in the society concerned (with classificatory relatives counting as kin just as clearly as those with a close genealogical connection).
This theoretical consensus suggests that it might be possible to base an explanation of endogamy on the idea of kinship identity and amity. But, before we do that, there is a major difficulty that has to be faced and overcome: the range of people covered by the sense of kinship amity is culturally variable, and in many societies it does not include relatives by marriage. Fortes makes the point in a particularly forceful way. Citing the widespread saying that 'we marry those whom we fight' he comments that 'It is as though marriage and warfare are thought of as two aspects of a single constellation the direct contrary of which is kinship and amity' (Fortes 2006, 234). He notes that 'The opposition of kinship and affinity is most conspicuous in structural arrangements, as well as in moral and jural norms, in societies with exogamous unilineal descent groups' including the Ghanaian Tallensi with whom he worked (Fortes 2006, 235).
Less dramatically, but consistently with this, Goody also notes that marriage in sub-Saharan Africa does not carry the connotations of social equivalence that it usually has in Mediterranean and Eurasian societies (Goody 1976, 101-103). It is all quite different from the solidarity-inducing relationships associated with endogamous marriage in the circum-Mediterranean area.
This leaves us with the question: why does the sense of kinship identity (and the solidarity it entails) exclude affines in sub-Saharan Africa, but include them in much of Eurasia? This is the crucial difference which makes endogamy a meaningful option in Mediterranean and Eurasian societies. In the next section we will argue thatdespite the limitations of the model in Table 1 the explanation does indeed have quite a lot to do with property.

Property, identity and work
Goody's original explanation of the causal pathway from the division of labour in column (2) to the marriage systems in column (4) of Table 1 centred on the distribution of property. The explanation that we are developing now depends on the distribution and implications of the sense kinship identity. But something like Goody's proposed causal connection might still apply if the distributions of property and identity were closely connected. It would work best if property and identity were actually equivalent. Although this idea is not included in Goody's formal model, or directly stated anywhere else in PAR, it does seem to be implicit in much of Goody's discussion. In this section we will try to develop the idea in an explicit way, and work out some of its causal implications.
The basic thought would probably make sense to people in Carniaif not as an explicit statement, then at least as a guiding idea. When asked why houses were left to sons and not to daughters, an old man replied that sons carried on the family name. In another conversation, someone remarked that so-and-so's house (cjasa in the local speech) had come to endmeaning by this that none of his most recent descendants in the male line, who had been geographically quite scattered, had themselves fathered sons. Cjasa (like the Italian words casa and casato) links three ideas: the physical building, the group of people who live in it, and the male descent-line which issues from it (exactly in the same way as the Kurdish usage of mal, as house, household and descent group)thus referencing a sense of identity that is simultaneously social, biological and material, hence property based. This semantic combination, which is also reported by Just (2000) in Greece, seems to be characteristic of southern Europe.
The idea of an equivalence between property, identity and biological descent is also found in Africa. In fact it underlies the social meaning of bride-price. Although Goody discusses bride-price at length in Goody and Tambiah (1973), it is only mentioned briefly in PAR, where he writes that bride-price payments legitimate the status of children as members of the father's patrilineage (Goody 1976, 8). This remark is true, but somewhat understated. Ethnographies of African kinship unanimously stress that in patrilineal systems the function of bride-price payments is to cancel the potential identity of children with their mother's clan, and transfer it to the clan of their father. This is pithily expressed by the southern African saying that cattle (usually the main component of African brideprice payments) beget children (Comaroff 1980;Goody 1976;Kuper 1982;Tambiah 1989). The association of buildings with the property-based transmission of patrilineal identity is also found, since African homesteads are often built around the cattle byre (Kuper 1982).
Similar points can be made about dowry. Although Goody's discussion of the consequences of diverging devolution focuses on the property itself, and how it can be kept within the social group, one of the motives he offers for diverging devolution is the need to preserve the class status of the family by ensuring that its daughters marry well. In effect he is suggesting that property is being used to purchase social identity.
So it is very widely true that property conveys a sense of identity which can be combined with, or exchanged for, identities derived from the biological parents. But what exactly is the parental contribution to this combined identity? There are some indications that what makes a difference in this context is not the biological connection of shared substancewhich can perhaps be assumedbut the amount of energy or work that each parent contributes to the material well-being of the household.
In Carnia, the connection between property, work and kinship identity emerges from some of things that people say. It was considered very shameful for a man to move into the house of his wife's family. In such cases the man was said to have married 'in cuc'as a cuckoothe implication being that he was depositing his children in someone else's nest, when he should have provided the nest himself, either by building or paying for his own dwelling or by bringing his bride to live with him in his parental home. From this perspective, male work and property inheritance operate in the same wayas an affirmation that the new family takes its main identity from the husband.
The wife, who retains her own family identity, is referred to as living in the family of the husband. (Thus if Maria Timeus married Edoardo Gortan, she would be referred to as Maria Timeus in Gortan; the couple's children will have the family name Gortan.) The identity of the new family with the dwelling provided by the husband is very clear, but the family's social identity does not depend on his work alone. This is made clear by a popular saying that the 'wife is three corners of the house' ('la femina a je tre cjantons da cjasa'), which people explain by referring both to the importance of the wife's moral character and to the amount of work she does -Carnian women (much like senior woman in the household (kabani) in Sisin, Hakkari) were expected to make a full contribution to the agricultural labour on which the family depended for its subsistence. The houseas an identity-bearing unitdepends almost as much on her provisioning work, as on the property-providing work role of the man. The moral significance of work emerges also from the rhetoric of work as self-sacrifice, which both parents are prone to invoke as a basis for their authority over their children and grandchildren.
Both property and work are forms of material provision which carry with them implications of identity and authority. In the next section we will argue that this equivalence is not a cultural particularity of our own field sites, but reflects a universal principle which emerges just as clearly from the African data: namely that both forms of material provision are understood as giftsin the Maussian sense that they convey both identity and the expectation of solidarity (Mauss 1970(Mauss [1954). Though the principle is universal its implications varydepending on the economic roles of each sex, and the property transfers that take place between the generations and between the families of the bride and groom.
In the following section we will draw on these ideas, in an attempt to rework Goody's original argument on Maussian lines.

A Maussian reformulation of Goody's model
We will present the revised model in four parts. The first part defines some terms and assumptions. The second part deals with micro-level interactionsanalysing the transactions involved in each marriage, and the different patterns of property and identity which they can generate. The third part deals with macro-level analysis, looking at the implications of these same patterns for overall systems of kinship and social solidarity. The final part discusses ways in which these identity-based relationships interact with the economic and political considerations which also affect the transmission of property, the choice of marriage partners and the practical significance of collective kinship ties.

Material provision as a Maussian gift -Definitions and assumptions
We will start by defining our terms, and stating some assumptions. By material provision we mean any action that contributes to the material well-being of a person or group. The actions in question might be productive work, a transfer of property, or allowing the recipient to share an important resource, such as a house or land.
By gift we mean an action by a donor which benefits a recipient in some definite way and which also creates a social connection between the donor and recipient. This connection involves mutual solidarity, and creates a debt. One implication of this debt is that the recipient takes on (wholly or partly) the social identity of the donor (including collective identities such as clan membership). These relationships only hold for unreciprocated gifts. Repayment of the giftat the time or latereliminates the debt, and cancels the relationship between the donor and recipient.
We assume that all acts of material provision carried out by relatives are understood as gifts. Donors and recipients can be individual people, but they can also be social units such as conjugal families or lineages. The transfers of identity that result from gifts do not just affect the individuals or units directly involved, but also their relationships to others who share the same collective identities.

Micro-relationships: The triadic transactions involved in each marriage
We now apply these definitions to some social choices that are relevant for Goody's argument. The first of these choicesbetween patrilineal or matrilineal successionis not dealt with directly in PAR; but it forms the background to much of the discussion, which implicitly assumes patrilineal succession. We therefore include it here.
Any marriage involves three conjugal families: the new family of reproduction formed by the union of the husband and wife, and the separate families of origin of the husband and of the wife. In nearly all kinship systems the new family is more closely associated with just one of the original familiesinheriting its family name, or some similar symbolic marker, which shows that they share a common lineage identity. We argue that the sense of shared lineage identity follows from the gift relationships which are set up by the processes of material provision.
Where provision comes mainly from the husband's sidein the form either of the husband's own productive work or of property provided by his parents or relativesthe new family will be most strongly associated with his family of origin, resulting in patriliny. On the other hand, if the provision by the husband and his relatives is outweighed by the material contribution provided by the wife and her relativesthe debt to the wife's side will be greater, resulting in the matrilineal transmission of social identity. The crucial point is that the direction of lineage affiliation depends on the relative strength of the gift relationships with the respective families of origin.
We now come to a point which is central to Goody's thesis in PAR, and to our Maussian reformulation of his argument: the famous phenomenon of diverging devolution, the fact that intergenerational transfers of property do not always coincide with the relationship of named lineage succession. We extend Goody's discussion of diverging devolution in three ways: by including the other aspect of material provisionnamely the work of the married partners; by including property transfers to, as well as from, the family of the inmarrying spouse (i.e. bride-price payments in case of patrilineal societies); and by taking full account of the accompanying transmission of inherited identities (which Goody seems to have had in mind, but largely omitted from the discussion of the model in Table 1).
The key point is that, even if the inherited property was transmitted in a non-diverging way (i.e. entirely from the husband's family of origin, in the case of patrilineal succession) there will still be a debt to the wife's original family, on account of her productive work, which means that the new family will also be partly identified with her family of origin and therefore (despite inheriting the paternal name) the social identity of the new family will actually be somewhat mixed. In other words, even where property is inherited unilinearly, the inheritance of identity will diverge to some extent-bringing with it an ongoing connection between the lineages of the husband and wife. This is the case in our own examples of Hakkari and Carnia.
However, it is possible to ensure the purely patrilineal, non-diverging inheritance of identity if the husband (or his original family) pays enough bride-price to compensate the wife's original family for the value of her productive contribution to the new family unitthus effectively removing any identity connection between affinal relatives. Something like this appears to fit the African ethnographic accounts that we cited earlier. Payments of dowry, from the bride's side to the groom's side, would have the opposite effectstrengthening affinal ties and further weakening the sense of patrilineal identity. This is the situation that Goody considers characteristic of Mediterranean societies and of Eurasia in general.
(The absence of dowry in Hakkari and the minimal level of dowry payments in Carnia may mean that affinal ties are weaker in Hakkari and Carnia than in Mediterranean communities in which dowry is paid. But another interpretation would be that women's labouring role is more pronounced in these highland communities than elsewhere, 13 and that this effectively compensates for the lack of dowry. The key point for social and political organization is that, in both highland and lowland Mediterranean societies, the separate or combined effects of the wife's provider role and of any dowry payments is to ensure that families inherit some of their social identity from the wife's side of the family.) In emphasizing the inheritance of identity we do not mean to downplay the importance of the material transactions for the individuals and families concerned, or to deny that environmental conditions and the gender implications of particular technologies may play some role in setting the pattern of property transmission. But it is clear that there is a good deal of free-play in the system, and that the resulting patterns of lineal or diverging identity owe as much to the social arrangements of property transmission as to technical constraints on the gendered division of labour.

Macro-level relationships: Marriage and the structuring of kin-based communities
The ways in which identity is transmitted have major implications for the structure of collective identities and the pattern of marriages in society as a whole. Lineal transmission of identity (for instance by patrilineal inheritance of property combined with bride-price payments which repay the value of the wife's productive work) would be compatible with a situation in which the society is permanently divided into a set of exogamous unilineal clans which, although they exchange marriage partners, do not share a common kinship identity. In such a system kinship amity could not be used as a basis for overall social cohesion or for shared loyalty to state structures.
Diverging devolution of identity (for instance in patrilineal systems where there are dowry payments or where wives make substantial productive contributions which are not cancelled out by bride-price payments) would not be compatible with a system of mutually independent lineages, since each marriage would result in a blurring of the original identities, and to cross-cutting loyalties which would undermine the practical effectiveness of unilineal ties. However, there are two ways in which this blurring of ties could be avoided. One would be to marry endogamously within the clan or local group since then both the husband and wife would share the same identities. The other would be to focus the sense of collective identity on the society as a wholewhich would have the same effect.
Because they involve unions between partners who are in some respect alike, both these strategies can be described as homogamous. Though they differ in the genealogical and spatial range of the marriage ties they create, both strategies would result in social fields without any conspicuous divide between kin and affinesin which the rhetoric of kinship amity could be used to support coordinated action by the community as a whole. They can be contrasted with the allogamous systems created by the existence of mutually independent exogamous clans which, regardless of the geographic scale, always result in social fields that are divided between kin and potential affinesin which the latter are socially defined as strangers, with whom the ties of solidarity, if any, are correspondingly weak.

Identity and strategy
The previous subsections have described how social identities are shaped by, and also help to determine, the choice of marriage partners and the property and labour transactions involved in setting up new families. But as we have seen, those same transactionsof marriage, labour provision and property transmissioncan also be analyzed in terms of economic and political strategies designed to maximize the material prosperity and status of the people involved. Although, in the theoretical literature these perspectives are sometimes presented as mutually exclusive, the fact that both have received solid empirical support suggests that in many (perhaps all) societies it is common for local people to adopt both perspectiveswith the particular emphasis depending on the social context involved. The crucial problem, both for theorists and for local people themselves, is not how to choose between strategic and identity-based perspectives, but how to integrate them, so that social choices make sense from both perspectives.
There are two main points at which the identity-based and strategic perspectives intersect and influence each other.
One concerns the payments involved in setting up new families. Bride-price can be seen both as cancelling the affinal link (in an identity perspective), and as a payment made in exchange for the wife's future productive contribution (in Boserup's economic version of the strategic perspective). Similarly diverging devolution can be seen both as an affirmation of the identity connection between kin and affines (in an identity perspective) and as a strategic payment (by the bride's family) in exchange for the economic support that she will receive from her future husband. It is likely that the system as a whole will be most stable when the payments make sense from both points of view. This would tend to produce the correspondences between economic gender roles, marriage transactions, and marriage patterns set out in Table 3. The other point at which identity and strategy intersect concerns the form taken by homogamous marriagewhether or not its solidarity building potential is used to enhance the cohesion of local communities, leading to local endogamy in the full sense; whether a particular need for male solidarity leads to patrilineal endogamy; or whether it operates horizontally, dividing society as a whole into distinct social classes. These issues loomed large in the ethnographic section of this paper. They take us beyond the formal simplicities of theoretical model building, and pose the vital question of whether, and how far, our Maussian reworking of Goody's model helps to account for the historical developments and geographic differences that he was trying to explain.

Applying and evaluating the revised model
Rethinking the historical sequence Goody intended the model set out in Table 1 to reflect both the causal order of the variables concerned, and the temporal sequence in which they appeared. It depicted a movement which started with the advent of advanced agriculture, and then proceeded in a topdown fashion through the advent of state structures, the implementation of these methods in male agriculture, the consequent adjustments to property transmission leading to daughters receiving a substantial share, and finally the adoption of in-marriage to prevent this female share being lost to the social group. Table 4 sets out a revised version of the temporal sequence which would be consistent with our own data and with the revised 'Maussian' causal model which we developed in the previous section. It is not a complete break: as with Table 1, social arrangements are seen to respond to economic changes. But there are major differences. For one thing, the order of events is different. Table 4 starts small, with a change in gendered work patterns, leading on to a system of endogamous local communities (recorded in column 3), which sets the scene for a cumulative process of expansion and economic intensification (in column 4), which eventuates in a system of hierarchical states and a tendency to class endogamy (in column 5).
This process of cumulative development is surely more realistic than the sequence set out in Table 1. But it is not just the order which differs: the mechanisms differ as well. Both tables feature a sequence leading from male farming, through diverging devolution to inmarriage (columns 2-4 in Table 1, and columns 1-3 in Table 4); and in both tables the first step in the sequence (from male farming to devolution) depends on the pragmatic Maussian version what matters is that the shift in property transmission leads to the diverging devolution of identity, which opens the way to homogamous marriage and a general pattern of cooperation based on a common kinship identity. The fact that this leads to marriages within the same, or with nearby, local communities is not due to a strategy of hoarding property, but because these are the people with whom cooperation is important within what are still small-scale economies which depend on the shared use of common local resources. We have taken our own field sites as examples of the social processes involved, showing that they channel social competition towards economic development and hence to the kind of intensification processes that would in the past have led to the processes of economic and social change depicted in columns 4 and 5. Indeed the need (as they saw it) to defend the local community from the risk of absorption into the wider society, and from the development of class stratification were among the reasons which local people gave for endogamous marriage and for the avoidance of marriage payments. In Carnia this process of socio-economic transition was already far advanced at the time of fieldwork.
The reason was, as we have seen, that the economic basis of Carnian agriculture had collapsed. As this example shows, there is no fundamental conflict between a Maussian property-and-identity perspective and a Goody-Boserup economic-strategy approach, once we realize that inheritance and marriage transactionsand the strategic calculations which guide themmust always take account of both aspects. We believe that a combined analytic approach has the best chance of making sense of the political economy of kin relationships in particular societies. It may also provide the best prospect for understanding the (pre)historical processes that led to the establishment of class-stratified state societies.

The need for empirical validation
Our decision to base an argument about intercontinental differences and long-term historical trends very largely on data from two particular highland societies (which are themselves peripheral to the regionscircum-Mediterranean and Eurasianwhich we have used them to represent) clearly raises questions about the validity of our conclusions. What right have we to develop the theoretical implications of two case studies of endogamy (or as we would now prefer to write 'locally focused homogamy') in this way? In mitigation we would argue that we are hardly the first anthropologists to argue from the particular to the general, and that we have drawn extensively on other ethnographic data, as well as on theoretical works.
Ultimately, though, the value of conclusions derived from special cases can only be heuristic. Inference about large regions, or historical periodslet alone claims concerning universal mechanisms, such as those we make in our Maussian modelmust be based as far as possible on data that is fully representative. That is why PAR's statistical analysis of Ethnographic Atlas data is so important, and the question that arises is whether we can test our own conclusions in the same way. We think that we can.
If the arguments presented here are correct, we would actually expect a better fit to the statistical data than was achieved by Goody's statistical model. The improvement should show up particularly strongly if the statistical variables were modified to reflect the theoretical discussion in the last few paragraphs. (Relevant changes would be: to allow for the presence or absence of bride-price in defining the diverging inheritance variable so that it reflects the inheritance of identity; and to clarify the difference between homogamy and endogamy (and their respective contraries, allogamy and exogamy) since Goody's 'in-marriage' variable confounds the two conceptual dimensions (see Appendix 1, Table A1). It might also be possible to add statistical checks for associations between inheritance and marriage strategies and measures of conflictuality and violence). We plan to carry out these analyses, and to present our findings in a later article.

Conclusion -Goody and the mechanics of social reproduction
As well as being a big-picture man, Goody was interested in the detailed mechanics of social reproduction and organization; and what makes PAR almost unique in anthropological literature is the way it combines these two perspectives. Much of what Goody wrote in PAR would have been consistent with the argument developed here. The difference is in the way he specified the mechanics of the system. He saw these as driven by competition for property and social status. We have added the dimension of social identity and the deliberate creation of amity, arguing on Maussian lines that practical interactions generate relationships of identity, which feed back into the socio-political relations of production. In other words, we see identity as part of the mechanics of the system. Why did Goody himself not take this view? We agree with Hann that his 'resolute materialism' had a lot to do with it. But even more important was the motive which underlay his materialism. This was his search for the processes of causation which generated and distinguished different social systems. Though not himself a Marxist, Goody had absorbed the Marxian infrastructure/superstructure framework (Table 1 simply lays this vertical ordering on its side). This view was widely shared, and the idea that kinship identities and marriage exchange were part of the ideological superstructure influenced the way they were analyzed by other authors writing at about that time (Bloch 1975;Bloch and Sperber 2004;Parry 1986). The implicit assumption was that, although ideology had real effects on people's lives, it was not the moving force which shaped and transformed societies.
Since then a Geertzian view of kinship has put identity centre-stage, but has seen the transactions involved as systems of almost pure meaning (Carsten 2000;Comaroff 1980). Writers in this 'new kinship' tradition have emphasised the importance for identity of most of the elements that we have considered here, including the house (Carsten and Hugh-Jones 1995), ideas about agricultural production, biological relatedness and the work of nurturing (Bamford and Leach 2009;Carsten 1997), and the exchangeability of property and biological descent (Strathern 1990). However, because they do not look at the economic determinants and consequences of these meaningful interactions, they are unable to explain the distribution of the systems they describe over space and time.
The Maussian approach developed here draws these identity-transactions back into the system of causal relationships. We think that Goody would have approved.

Notes
1. With Eurasia we mean the entire landmass of Europe (including the circum-Mediterranean region of north Africa) and Asia since the urban revolutions of the Bronze Age, in the sense Goody (2010) has been using the term.
2. Table 1 is based on the diagram on page 38 of PAR. We have omitted variables that are not directly relevant to the theme of this article. 3. Goody defines advanced agriculture as a condition which allows an individual to produce more than he/she can consume, be it through the use of plough or irrigation, being characteristic of Eurasian societies (1976, 20ff). 4. The number of societies included in the Ethnographic Atlas has been steadily growing in the decades since Goody wrote (Gray 1999). This is the number included in the version that Goody used. 5. These conclusions can also be deduced directly from the correlation coefficients between 'inmarriage' and the other four variables, shown in the final column of Table 2. The correlation coefficients with 'complex polity`and 'male farming' are about the same as the correlation with 'diverging devolution'. If the only link between these two variables and 'in-marriage' was the indirect one through 'diverging devolution', their coefficients would be much smaller. 6. The particular tribe mentioned here in the 1980s had been settled in a mountainous region of about 400 sq.km. and a rough estimate of the tribe's population was around 3-4 thousand people (Yalçın-Heckmann 2000). 7. The Hakkari Kurds have categories for relatives by patrilineal descent (xwîn/waris) and also affines/matrilateral kin (xism). Because of the practice of FBD marriage (or more generally marriage within the patrilineal descent group), membership of these categories overlap. Both categories overlap with residence in the same village. In practice the preferred partner choice is with someone who is closely connected in one or more of these three ways (descent, affinity, residence). 8. Leach (1940, 19-20) describing endogamy within the patrilineal clan (taifa) notes that 'all [within the tribe] call one another "amoza"', a term for FB's children, the southern Kurdish equivalent of pismam. 9. Altuntek (2006) argues, for example, that among the Kurds in Eastern Anatolia women were very important for linking patrilineages and the relations to MBs were institutionalized. For similar arguments on the role of affinal relations in patrilineal societies, see Khuri 1970. 10. 'Generalized' in the sense that sons and daughters from any household can choose their partner from any othernot in Lévi-Strauss's sense of repeated asymmetric exchange between the same lineages. 11. Bride-price was a common practice in most of non-tribal, exogamous, and prestige marriages. 12. This is a point also made in early research on Kurds, especially by Leach (1940) and Barth (1954). 13. The equal participation by women that we have described contrasts with Ethnographic Atlas figures which show predominantly male farming in 65 percent of circum-Mediterranean societies (PAR page 131 Table 21). 14. The number of societies included in the Ethnographic Atlas has been steadily growing in the decades since Goody wrote (Gray 1999). This is the number included in the version that Goody used.