CHAPTER SIX Cognatic Descent and Ego-centred Groups i In Chapter One, we contrasted the unilineal and cognatic methods of recruitment with respect to the forming of descent groups. The unilineal method, we saw, had the advantage of assigning individuals^.one group only.,(father's or mother's) and so creating 3iscrele|ffl;i'at'is non-bverlap-ping groups. We can perhaps visualize this as below: matrilmeayts pairiiineacics A B C D E r- G H COGNATIC DESCENT 14.7 ter, although as we saw for the Yako, it might if not skilfully 'meshed' produce some problems. We might illustrate it as below: Diagram sg Diagram s8 When both unilineal principles are used, the same remains true and a man is a member of one rnatrilineage and one patrilineage and these exist for different purposes. Either unilineal system produces a discrete series of descent groups - lineages or clans - and an individual is assigned to one of these only. In a dAuble^djscent system, the units of the system are still discrete and non-overlapping patrilineages on the one hand and rnatrilineages on the other, although there will be members of all the rnatrilineages in each of the patrilineages and vice versa. As the malrilincai>es and patrilineages exist for different purposes, this does not mat- When we come to cognatic descent groups, however, the picture changes. No longer is the society composed of discrete non-overlapping groups, for as we saw, by their very nature the cognatic lineages are bound to overlap in membership, and a man will be a member of several similar-purpose groups at the same time. Clearly, then, these must present different structural problems from unilineal groups, although, like the latter, they are composed of die descendants of a common ancestor. In the cognatic case, however, this descent is not limited by sex, but all the ancestor's descendants are included in his group, ft represents the dlird alternative open to our sibling group: allow both men and women to reproduce the group. We might illustrate it as follows: I the opposite may be equally true. This sobering lacl should 156 KINSHIP AND MARRIAGE perhaps turn us from asking, simply, 'into what structural category -..liould vvc place system Xto asking,' what are the trends in system X; horn where is it coming and where dots it seem to be, going •'1 III We have seen, then, that cognatic descent groups can.be of three kinds: 1. UnresiricUd-'iii these, all the descendants of the ultimate or founding ancestor are members. •2. Restricted... In these, all the descendants of the founder have a right to membership, but can only exercise this right if they choose, say, to live in the founder's territory. 3. What we might call pragmatically restricted. In these all the descendants remain members, but in practice they cannot take up membership in all the groups they belong to, as these are territorial. So they have to choose which one to affiliate with, but this is not immutable. The important thing here is that the. restricted variety can, in (act, function with the same effectiveness as unilineal descent groups, and also have an added .flexibility that might turn out in some circumstances to be a positive advantage. Lei us look at some examples of these kinds of groups in action in order to see just what they can do. We have discussed the third type under the Maori, so we can leave thai aside for tin; moment. Lei us turn to the Gilbert Islanders in the Pacific to see an elaborate cognatic sysiem at work. The Gilbert Islanders1 have several kinds of kinship group but we will concern ourselves with their cognatic descent groups. The most all-inclusive of these is an unrestricted cognatic descent group known charmingly as the 00. Both men and women hoLd land and on the death of an individual his land is divided between all his children. (His daughters may have received their share on marriage.) As this process continues, a tract of land is divided and subdivided amongst the descendants of COGNATIC DESCENT I57 the original owner. The oo regards itself as in a sense jointly responsible for all the land, and members of the oo may not sell land without the permission of all the others. If any line of the oo dies out, then the land reverts to the oo generally and is redistributed among the members. Members who leave the area in which the descent group owns land, do not thereby lose rights in it. Any one who is descended from the original owner keeps his rights in the land and passes these on to his children. The fact that the oo are bound to overlap means that an individual may hold rights in several of them. In such a system the various plots of land that an individual holds in the various oo territories must not be too far from each other or he could not work them. On small islands, such a system oflandholding is feasible. The Maori hapu, on the other hand, were, not so compact, and so multiple, inheritance was relatively impossible. Another important descent group on the Gilbert Islands is the bmoli. This is a segment of an oo which is concerned with seating-rights in the community meeting-houses. These rights are very important to the Gilbertese. Each meeting house is marked out, and certain areas of it belong-to the descendants of men who owned particular plots of land. Now all the descendants of one of these men would be an oo, but not all would have inherited a piece of his land. When a man died his land would be divided amongst his children, and he would bequeath tire land in one of his oo to one child, that in another oo to another child . . . and so nu. Thus a child might be a membei of an oo but nut necessarily Slave inherited any of its property; hence he •wi/uli.l not be able to sit with the bwol.i associated with the oo. He. would, however, have got some land in at least one oo that had bwuti rights in one of the meeting houses. A person would so distribute his property to his heirs that each of them obtained such a right. The division of inheritance is such that men got much more than women, and in consequence a man is more likely to get bwoti membership from his father than his mother. This gives the hvoll a patrilineal tinge. Early writers often described it as a palrilincage. i58 KINSHIP AND MARRIAGE COGNATIG DESCENT The bwod is then a common descent group whose membership is restricted to those descendants of a common ancestor who have acquired rights in a particular plot of land. Thirdly, the Gilbertese have the kainga. Now, every ancestor who founds a bwoti, also founds a kainga, hut the membership rules are different, so that although each kainga is associated with a bwoti, their membership is not coterminous. The rule for kainga membership is again hitched to landholding. The original ancestor had lived on a certain tract of land. Some of his descendants continued to reside there but others moved away. Those who continued to reside there plus those who had been born and raised there but had moved away after marriage, formed the kainga. Thus, those who were born on the land inherited membership even if they moved away; but if they moved away their children did not inherit membership. Thus, if a man's parents were living patrilocally he would belong to his father's kainga: if they were living rnatrilocally he would belong to his mother's. It was thus in a sense parental residence choice that determined an individual's kainga membership. Since residence was predominantly patrilocal most people belonged to the father's kainga. Leadership of the kainga was passed on patrilincally. This was worked by having the eligible successor reside patrilocally so that his son would be eligible to succeed him and so on. 'I'bus, (he kainga very much resembled a patri-lincagc, but I his risen iblancc was arrived at by a route iar different from llie simple rule of patrilineal succession. ■ (joodenough, who describes this system sums it up thus: . . all three descent groups are somehow connected with land. An ancestor having established ownership of a tract was the founder of all three. All of his descendants formed an oo. Those in actual possession of a share in the land are eligible to membership in a bwoti. Those whose parents resided on it form a kainga.' Thus the oo is concerned with rights in the laud ; the bwoti with actual possession of a piece of it; and the kainga wi th residence on i t. Similar groups to these three - particularly the oo and the '59 bwoti — are found in the Philippines, the Solomon Islands, Samoa, Polynesia generally and, of course, New Zealand. The Scottish 'clan' was a form of cognadc descent group, claim in Gaelic simply meaning 'children' or descendants. Because of a preference lor endogamy and patrilocal residence it, too, had a strong patrilineal tinge as is evidenced in the inheritance of surnames. But note that every true Highland Scot bears two names: his father's and his mother's. Thus Robert McAlpine McKinnon is a McKin-non through his father and a McAlpine through his mother - and he may belong to other clans 'by birth'. This system no longer functions except for sentimental reasons, but in some of the more remote Celtic parts of the British Isles, a descent group like the oo, with very similar functions, is still in existence. It would be described with the old Gaelic word too — as 'Claim Eoghain' for example: die children or descendants of Owen, over as many as eight or more generations. At this point, the reader might like to refer back to Chapter three and look at case 5. In discussing the possible environmental pressures that might produce various kinds of grouping, I suggested that 'transferability of skills' in males, and the need to distribute a population over agricultural plots might lead to a situation in which either the. men or the women moved on marriage, thus creating ambilocal residence. In the examples we have just looked at, we can see how this has happened in a number ol eases, 'the residential group that we would get in cast: 5 would be the. core of the Gilbertese kainga (with the spouses of members). If those members who were born there and subsequently moved away on marriage continued to have rights in the land of the group, then a true kainga could easily come into being. Such a combination of groups that wc get in the Gilberts and elsewhere on small islands, however, are very tied to plots ofland and localities. This is fine for small islands, but clearly it would be of little use to desert nomads or expanding- 'warrior tribes. For these, a patrilineal system has obvious advantages and such a system of groups as we find i6o KINSHIP AND MARRIAGE COGNATIC DESCENT 1G1 on the. Gilberts would not be workable. They represent a lather fine adjustment to ecological pressures: an exploitation of our third option which human bioiogy offers for tbe recruilment of groups on a kinship basis. Before leaving cognatic descent gioups, we should perhaps look at one for which case 5 in Chapter three can serve as the exact model. The Iban of Borneo live in Tonghouses' and each longhouse may contain anything tip to 50 families each living in its own apartment. These families, known as bilck alter the name of the apartment, are the real units of the socirty, rather than the longhouse as such. When a couple marry, they must decide whether to live in the bilek of the man or that of the woman. This is a momentous choice, because they become members of the one they live in and lose membership in the one they turn down, if a man chooses to go to his wife's bilek and farm the land that it owns (part of die longhouse land), then he loses all rights to the land of his natal bilck and is thoroughly incorporated into his wife's. ('J'he bilek family is cxoganious.) A bilek family consists, then, of all (lu: descendants of die original owner of die apartment, except (hose that have moved away and their children, hut includes the spouses of natal members who continue to reside in it. Piece an individual does not choose whether or iiol to Jive in the mother's or father's group, but in the natal group (whether this be mother's or father's), or the spouse's group, It is a kind of cognatic version of the Chinese practice of incorporating the wife of a man into his lineage. It is interesting because viriloral and uxotiloca! residence are equally balanced in Iban society, and it therefore gives us an example of cognatic recruitment which is not predominantly patrilocal. The Iban arc rice farmers, and one rice farm is very much like another, so labour and skills are easily transferable. A couple on marriage therefore must judge which is the best bet - his bilek or hers. Thus, we can see that out of a situation such as we envisaged in case 5, a number of possibilities could emerge. Predominantly putrijoeal residence and continuing rights in the land on (he part of those who left would lead to a Cilbertian situation, while a more rigidly residential rule of membership would give us Something like the bilek of the Iban. That such groups as the 00 and the kaingo seem confined to small islands is perhaps significant, but we should not lose sight of the Maori'- a numerous, sophisticated and warlike people -who show that a cognatic principle of descent-group formation need not be confined to atolls. I have approaches the problem of cognatic descent groups backwards in comparison with the approach used for unilineal groups. With the latter, f tried to show how the principle of unilineal descent-group organization could arise out of a simple residence situation: in the present case I started with the principle of group organization and ended with a rather tame reference to residence. The main reason for this was that I wanted to align these groups firmly with unilineal groups in the common category of descent groups. Hence, I started by exploring the possibility of having all the descendants of an ancestor as members of his group. This gives us a continuum:' at one end wc have the unrestricted cognatic descent group in which all the descendants ol the ancestor are members; then wc have cognatic descent groups restricted in membership in terms of residence; (hen we have descent groups restricted in terms of sex - that is, only allowing the members of one sex to recruit the group, 'thus unilineal groups are seen simply as one type of restricted descent group rather than as a completely separate type of group altogether from the cognatic. All these groups share . in being common-descent groups - their focal point is an ancestor from whom all the members ultimately trace descent. The second reason that I approached this problem backwards was because I am less sure of the connexion between residence and descent in this case. As we saw earlier, tire cognatic descent group seems to be compatible with any kind of residence principle - it really depends on what the purpose of the group is. Where, as is usually the case, it is concerned with the inheritance and control of land, then perhaps we can more easily sec that residence on die land might have something to do with it. The circumstances that 162 KINSHIP AND MAliRIAQJi COGNATIC! DEbCENT J 63 favour the development of ambilocal residence (case 5) might well favour the growth of cognatic descent groups. Given the transferability ol skills and a subsistence-agriculture economy with a pressure on land, then a system of cognatic descent groups would provide a reasonable solution in that it allowed for a redistribution of population amongst the scarce plots. A system like the Gilbertcse is an admirable answer to the problem. The unrestricted descent group operates in relation to claims on the founder's plat, while the restricted descent groups operate with respect to the actual ownership of parts of it, or residence on it. Such a system can have its problems of course, not least that ol fragmentation, and the product may be small, scattered holdings which are uneconomical and difficult to farm. If residence is ambilocal, and the areas of land concerned are distinct, then the members and their children who leave can either retain rights in the land or lose them, thus producing unrestricted or restricted descent groups. If, however, the people live in large settlements and not on their land, and residence rules are flexible, how then do such groups arise? The lull set of determinants for these groups has not yet been worked out, and it may be that an independent ideology of the equal rights of all children to inheritance is involved. But it is hard to believe that the ideology would survive in the face of environmental pressures that made it nomadaptational. Ideological factors cannot be ruled out, however, because it is possible - with suitable adjustments - for unilineal systems to survive in much the same circumstances as seem ti> breed cognabc systems. Ecology sometimes sets hard and fast limits, but very often it allows a large amount of'play', and so different systems can flourish in the same conditions. But we must not forget the theorists who insist that cognatic systems arc breakdowns of unilineal systems in the face of environmental pressures. Thus they may represent an adjustment of a unilineal system. Alternatively they may simply be the breeding ground of unilineal systems. On this subject, we have a long way yet to go.. My reason for wanting to put these cognatic groups firmly into the category of 'descent groups' is largely due to the fact that anthropologists have tended to ignore them until recently. Most students of kinship, following Raddiffe-lirowii; have been bemused by the hard, clean beauty of the unilineal principle and have seen in such things as the hapu only sports and oddities. They have either tried to assimilate them to unilineal systems, or have just ignored their existence. Thus, the 'descent-group theory' that gets much talked about really means 'unilineal descent-group theory'. Radcliffe-Brown thought the advantages of the unilineal solutions to be so obvious that he could hardly imagine how any society could get by without adopting one or lire other of them. Quite a number, however, have managed to stagger along despite this handicap, and we are now becoming better equipped to see why and how. There has also been another confusion. Anthropologists have thought that the cognatic principle could not be effec-dvely used to form descent groups - those based on descent from a common ancestor - and have thought that it was solely concerned with the formation of ego-centred or personal groups. It is to these 'hr.i- we must now turn. IV Descent groups have certain characteristics in common whatever their form. They all consist of the descendants of a common ancestor; all the members are therefore related to each other in respect of such descent. They are usually 'corporate' groups, that is, groups that exist' independently of the individuals composing them. They exist 'in perpetuity'; individual members come and go, but the group goes on. Corporateness also implies that they act 'as a body'; thus if one'of' their' members.kills a man, the group, as a whole is held responsible.for the killing; or, as is often, the case with land, this cannot be alienated by an individual member but belongs to the group as a whole and must only pass from one member of the group to another. Descent groups are not always corporate in this latter sense, but they always are groups 164 KINSHIP AND MARRIAGE that exist in perpetuity. They .are.commonly .exqgamous.,, but this is not universally the case. It is the relation by common descent of all the members to each other, andcorporate-ness in the sense of perpetuity of existence that characterizes all descent groups. Bearing this in mind, we can now look at ego-centricity. We saw earlier that there were two ways of looking at any kinship system: from the angle of the kin-groups composing the society and from the angle of the individual and his kin. Thus, we saw that looked at from the first angle we may only see a society composed of patricians; but that in such a society an individual may recognize cognates up to a certain degree as relatives, and have important relationships with matrilateral relatives . . . and so on. Now, this is true of all kinship systems. Goodenough has christened..,thgse-.-two angles the ancestor-focus and the ego-focus. Now, while all kinship systems can be viewed from either locus, only some make use of the ancestor-locusjajUjeJ^r^^ -descent groups; others make use of the ego-focus in group ibrmatron and "it is this formation of groups on the basis of the ego-focus that we must now look at. Let. us note that these are not mutually exclusive methods, and a society can have more than one kind of kinship group operating in. it. Cm-pups formed on the ego-ibcus must, of necessity, be.yery different from those based on the ancestor-focus. They consist not necessarily of people who have an ancestor in common, but of people who have a relative (ego) in common who is not an ancestor of theirs. The best known of such groups is die hlmliul. This is recruited on the basis of the degree of relationship of its members to a common ego rather than a common ancestor. The best way to illustrate this is by the 'familiar English notion of cousinship. Thus, all ego's cognates up to, say, second cousins, could be counted as his kindred. Diagram 32 illustrates this by using the neutral square to mean 'person(s) of either sex' - which stresses the cognatic nature of the group. But it is very different from a cognatic descent group. The men bom of the kindred are; not all related to each whereas they are all related to ego. COGNATIC DESCENT 165 EGO(S) Diagram 32 All the members of it do not have an ancestor in common; all they have in common is ego himself (or herself). Thus, every person in the society has such a group, and each group is relative to that person. No two people except siblings .will have the same kindred, and kindreds will thus endlessly overlap. We can illustrate this as below: Diagram S3 (solid line = I's kindred; dotted line = Id's kindred) Here we have a simple kindred of first-cousin range. II and III are members of I's kindred; I and IV are members of IPs kindred; but IV is not a member of I's kindred . . . and so on. If we carry the analysis lineally - over the generations - then we would find that ego's kindred was different from the kindred of his father, and that of his mother. Such groups clearly then cannot function except in relation to the ego who is their ioeus. They cannot be 'corporate' m l66 KINSHIP AND MARRIAGE the sense of existing in perpetuity, because once ego dies (and here siblings are counted as a collective ego), the group ceases to exist. This is not true of a descent group. Nor do ego's children inherit his kindred. The kindred then is a purely personal group. It is easy to see that such a group cannot perform the same functions as a descent group. It cannot be a landhold-ing corporation passing on land to its descendants. It cannot be in any sense a 'constituent' unit of the society, because it comes in and out of existence as its focal egos are born and die. What does it do then and how does it work? Well clearly it would be a useful group to have in societies where people operate independently, but need on occasion to call in help for some purposes. The Iban, for example, have made the bilek family their domestic and economic unit. They lack any form of descent group that is more inclusive than the bilek, and the longhouse is not a corporate unit. But the Iban put out quite spectacular raiding and trading parties of considerable size. These are recruited by means of the kindred principle. The Iban surround ego with a kindred of up to second cousin range. Thus each Iban has a body of people - all those related to him up to the degree of second cousin - on whom he can call for some services to himself and who have some obligations to him. He himself of course is a member of several such kindreds -those of his first and second cousins. Now when an Iban wants to take out a head-hunting party, he calls on the members of his kindred. They in turn can call on the members of their kindreds who are not members of the original ego's kindred, who in torn can call on the members of their kindreds ... and so on until the requisite number of men are mustered. Thus, in our diagram 33, (assuming that these are second-cousin range kindreds rather than first) I would call on II and III; II would call upon IV, who would in turn call on his kindred mates other than II .. and so on. This body would then go on the hunt and share the spoils between them. It is also possible to make ego's kin in some degree res- COGNATIC DESCENT 167 ponsible for him. Thus, in the payment of blood money, it could well be the kindred that was the operative unit rather than the clan or lineage, if a man killed another, then all his kindred would have to pay out blood money to the kindred of the dead man who would share it between them. Amongst the ancient Teutons this is supposed to have been the case, with the nearest kin to the murderer paying most, and the nearest kin of the dead man receiving most. In some systems (England under King Alfred for example), the pa trilateral relatives paid and received more than the rnalrilateral. The kindred here was known as the sib -a word that has been wrongly appropriated by some writers ior application "to unilineal descent groups. Amongst the Teutonic peoples, the sib was the exogamic unit, and this method of fixing the degree within which marriage was forbidden was adopted by the Christian church. The kindred could also be used for purposes of inheritance, even if it could not be a property-owning group itself. Thus, if a man died without heirs, his land could revert to ins kindred lor distribution amongst its members - perhaps again on the basis of'nearness'. The essence of the kindred then is that all ego's cognates up to a certain degree are recognized as having some duties towards him and some claims on him. It is pet haps wrong to call this a 'group' at all, but rather should we call it a 'category' of persons. It is never a residential unit nor is it corporate, and it only conies to life, as it were, when the purpose for which it exists arises - like headhunting or the payment of blood money, or the regulation of marriage. (In the latter case it need not exist as a group at all. AH ego need know is that he must not marry within a certain degree of relationship.) It is, then, a category out of which a group can be rccruilcd by ego for some purpose's. If we look back to tfie Gilbert Islanders, we will find that among their kin-groups they have, in addition to the oo, bwoii and kaiuga, a kindred called the iduii. Their Malayo-Polynesian relatives in the Northern Philippines whom we have mentioned as having cognatic descent groups, also 168 KINSHIP AND MARRIAGE cognatic; descent 169 combine these with personal kindreds of third-cousin range- The descent groups regulate ceremony and the use of land sites; the kindreds deal with homicide payments and regulate exogamy. They also come to the aid of an individual in trouble, but because of the overlapping of kindreds, this is only really effective, if the trouble is between two people so distantly related that their kindreds do not overlap. Clearly, if the two kindreds do overlap then some members will have divided loyalties as they wiil be equally members of tire kindreds of the two combatants. This, in fact, can be quite elfcctive as these 'overlap' members will then make strenuous efforts to bring about a settlement. There are many other examples of the co-existence of descent groups anel kindreds, each serving different social purposes. Kindreds can and do co-exist quite easily with unilineal descent groups too, but we do not need elaborate examples of this to see how it could work. I must again stress that the kindred is not really a group in the sociological sense. The fact that amongst the patrilineal Zulu a man may not marry any woman descended from ftis great-grandparents established that each Zulu has an exogamous kindred of second-cousin range. But that is all. Nothing else follows from this, and the kindred has no other functions. We should perhaps clear up one point that has caused some confusion. Que way that the kindred was reckoned amongst the Teutons, and one way that it can always be calculated is in terms of stocks. Now a stock is all the descendants of a person or of a married pair. Thus, a kindred of second-cousin range such as we have drawn tin diagram 32, will consist of lour .stocks - the descendants of ego's four pairs of great-grandparents. (A, B, C, and "0 on our diagram.) A kindred of third-cousin range, such as that of the northern Philippines, would consist of eight stocks ... and so on. Now, the trouble with the definition of the stock is that it is the same as the definition of cognatic lineage, and this causes confusion. Some writers have called the 'stocks' of the Teutonic sib 'non-unilinear cfe.sc.cnt groups' for example. The reader should be able to see what the confusion is here. The essential difference is of course that the cog- ^ natic lineage, in common with other descent groups, is "~s founded at a point in time and persists over time from then _j > on; the stock of a kindred exists only in relation to a par tic- V ^ ular ego and it disappears when he dies. If a member of a y- -| cognatic lineage dies, the lineage still continues; when the focal ego of a kindred dies, then the stocks are no mote. The lineage then is defined relative to an ancestor who H~ remains a fixed point of reference; the stocks of a kindred are «»4. defined relative to an ego. The stocks of a kindred the.)! arc, like cognatic lineages, all the descendant of a person (or couple); but unlike cognatic lineages they are not inde- * ; pendant entities, but only part of the circle of kin around art ego. Thus a cognatic lineage is a stock, but a stock is not N; necessarily a cognatic lineage, and when it is simply a '< constituent of a kindred it is really nothing like such a ^ lineage. < X.- ^ We have concentrated above on die cognatic kindred. Indeed I have not yet bothered to mention that there is any other form. I wanted to deai with this form of the kindred first, because of the confusion that has arisen in anthropology from dividing the world into societies with unilineal kinship avid those with cognatic, and assuming that the. only form of kinship organization compatible with the fatter was the personal cognatic kindred. We have seen that what matters is not so much the division into unilineal and cogitatic, as the difference between the cgo-juau on the one hand with, its personal ...'groups.'., and. the ancisioi-Jocus on theother .with its descent groups. We can clinch this by showing that other fori ns of'personal kindred exist than .the cognatic - forms which employ a unilineal principle in recruitment, if we wse_.uiiiHneaI_.as synonymous with 'unisexual'. The kindred can be broadly defined as 'ego's relatives up to a certain fixed degree'. What matters is how this 'degree5 is defined. It need not be defined r.ognaticaUy (or 'bilaterally' as it is usually called in the literature). The Kahnuk X:- i4 L u IJO KINSHIP AND MARRIAGE Mongols, for example, have a personal kindred consisting of all the people related to ego through males within a fixed degree. This is illustrated below. Diagram 34 Say the fixed degree involved is that of second cousinship -all the descendants through males of his great-grandfather (father's father's father). Then ego's kindred would be those people' on the diagram who are shaded. You might immediately object that this is a lineage - all the descendants through males of a common ancestor. So it is, but like the stock of a kindred it is not a lineage decided upon in terms of descent from an ancestor, but in terms of the degree of the relationship of its members to ego. Thus, all the people on the diagram are the descendants of a common ancestor, but they are not all members ol ego's kindred. Those in white however are members of ego's father's kindred. Tims the point of reference for membership is ego and not an ancestor as a fixed point of reference. To confuse this group - known as a patrilateral kindred. - with a patrilineage is to fall into the same error as we discovered in the confusing of cognatic lineages, with the stocks of a cognatic kindred. The Kalmuk case should underline the fact that the real distinction is between the two foci -- ego and ancestor: between decent groups and personal groups. Even when ego-centred groups recruit unisexually they are still more like COGNATIC DESCENT I'/l cognatic kindreds than they are like unilineal lineages. The lact that they recruit unisexually has some important effects that makes them different from cognatic kindreds, it is true, but it does not turn them into lineages. As long- as a recognized category of persons exists in the society which has as its point of reference an ego to whom all the category are related, then whatever its composition such a group will be of the. kindred type, even if it is not symmetrical and cognatic. Thus, on the island of Truk in. the Pacific, where the corporate units are matrilincages, each ego is surrounded by a group of relatives, which is named, and composed thus: 1. Ego's grandparents and grandchildren 2. The members of his matrilineage 3. The members of his father's matrilineage 4. The children and grandchildren of the members of his matrilineage 5. The children and grandchildren of the members of his father's matrilineage. This group, as I said, has a name, and certain lights and duties towards ego; it is constant only for siblings, and its membership is fixed by degree of relationship to ego. Here it is not so much a case of an ego-centred group co-existing with unilineal grouping: the ego-centred group absorbs ego's unilineal groups. In a number of unilineal societies, such clusterings of kin around ego exist, but they are not always by any means formalized and given a name and tin lies towards ego. This example stresses then the diUeivuce between analysing a kinship system from the cgo-i'ocus ~ which can be done ior any kinship system - and the system itself using the ego-locus as a means of forming groups or catego2'ies of kin for various social purposes. There is a great deal more that we could go into here, but space and probably the reader's patience forbid. Enough perhaps to have grasped these essential points: 1. That the division between groups descended from an ancestor and groups based on degree of relationship to an ego is fundamental. 172 KINSHIP AND MARRIAGE" 2. That both kinds of group can recruit either cognatically or unisexually. 3. That these modes of grouping are not mutually exclusive and can co-exist in one society serving different purposes. The fact that two otherwise different systems both use, say, the cognatic principle of recruitment, is important and makes it worth comparing thern. But it should not lead us to lump them together on this one criteiion. To help clarify this point Í offer diagram 35. Here the intersection of two factors - locus, and mode of recruitment - gives us our types of grouping. Focus Recruitment ego ancestor Unrestricted cognatic kindred unrestricted cognatic lineage Restricted 'unilateral' kindred unilineal lineage by sex other ? restricted cognatic lineage Diagram 33 The blank cell could be filled by an example of an ego-centred group restricted on the basis of residence with ego, although I know of none at the moment. Obviously the system of cognatic kindreds rings a bell for most readers as it resembles our own kinship system which is however, unformalized and lacks named kindreds. We simply recognize that relative.? on both 'sides' of the family are our kin, and we may interact with these, invite them to ceremonies etc. Unless the personal kin-group is formalized in some way, it is perhaps better to speak simply COGNATIC DESCENT 1/3 of ego's kinship network, and to spell out its form and functions. Our own system is primarily concerned with the nuclear family as its basic unit, and continuity over time is not of great importance. At ego's marriage two families are linked - his own and his wife's. The family he is born into is sometimes called by sociologists his 'family of orientation5, (a barbarous usage - 'disorientation5 might be more appropriate in many cases). The family he forms at marriage is his 'family of procreation' (very ambiguous but now accepted). Thus, our' kindred' cubists of linked nuclear families - ego's family of orientation, his family of procreation, his wile's family of orientation, the families of procreation of his siblings and children, and so on. Diagram 36 The limit of recognition of'nuclear-family' linkage tends to be narrow. This system is more reminiscent of the Shoshone or some Eskimo than of the more elaborate systems we have been discussing here. Cognatic descent groups can form on an ad hoc basis if property is involved, but there is not, above 174 kinship and marriage the level of the family, any extended kinship group to which people must belong. If such groups arise, it is to meet specific eventualities; they are not constituent units of the social CHAPTER SEVEN structure with legal status. Exogamy and Direct Exchange i After the uphill struggle of the last chapter it will perhaps be refreshing to turn back to the fundamental topic of exogamy; a topic we have been taking for granted up to now. We saw in Chapter two how prohibitions on incest automatically produced exogamy — because of the association of sex and marriage - but that the reverse was not necessarily true. In consequence, we could not accept that all exogamic regulations were simply 'extensions' of incest prohibitions. We saw in the subsequent chapters how exogamy presented a problem to those descent groups that practised it, because it made them look outside themselves for brides, and so forced mem into relationships with other descent groups. Now 'forced' here is simply a figurative way of speaking about the situation, and it may be misleading. Why should the descent groups not actively want to marry women other than-their own? In many cases they will not think too consciously about it; the rule of exogamy, whether it applies to lineage or chin or botli, is, like the incest taboo, a part of cultural inheritance. But, unlike the incest taboo, its benefits are more obvious to the people practising it, and they can often verbalize these quite cogently. It may here be the case that the continuing benefits of the rale may in fact be closely connected with its origin - which was not the case with the incest taboo. Like a good preacher, I will offer you a text for my sermon on exogamy: 'Then will we give our daughters unto you, and we will take your daughters to us, and we will dwell with you, and we will become one people.' (Genesis 34: r6.) I am not original in offering this text; the anthropologist