channeling the flow of loyalty, delineating the direction of alliance. It is a âmodel of,â a ...... considered solely the business of the individual smaller lineages initially involved. Second, conceptual ..... School of Oriental and African Studies, University of London. Thanks are due to ... Manchester : Manchester University. Press.
Does Complementary Opposition Exist? PHILIP CARL SALZMAN McGill University Uncritical application of lineage theory can obscure the complex reality of political organization and process in segmentary societies. However, the argument that lineage theory i s a folk ideology with no basis in behavioral reality seems unsupportable in the light o f comparative and historical evidence which indicates that short-term and long-term territorial stability o r instability are operative factors in the symbolic and behavioral significance o f lineage organization. [segmentary lineage systems, complementary opposition, ideology, political alliance, ecology]
I against my brother; I and my brother against my cousin; I and my brother and my cousin against the world. So the hoary saying of the Arab goes, describing the nature of past and current reality, providing an injunction for future action, expressing a core sentiment of Bedouin culture, channeling the flow of loyalty, delineating the direction of alliance. It is a “model of,” a statement about the way things are and the way things are expected t o be, and it is a “model for,” a statement about the way things should be and the way things must be if morality is to reign and chaos is to be avoided. This saying expresses a pattern of loyalty and alliance that social anthropologists have called “complementary opposition.” This pattern has drawn the most attention where it is, or is thought to be, the primary pattern of political structure and the primary process of political organization in the society. Societies where complementary opposition is primary are said to have segmentary lineage systems. Two elaborations on complementary opposition
PHILIP CARL SALZMAN. a McGill faculty member since 1968, studied anthropology at Antioch and Chicago (Ph.D. 1972). His field research has included the Chicago courts (1965) and Iranian Baluchistan (1967-68. 1972-73, 1976). emphasizing adaptation and social and political organization of nomadic tribes. He is organizing a commission (to be under the I.U.A.E.S.) which would contribute sound knowledge and proposals to discussions of the future of nomadic peoples.
Salzman, with Haji Nezar Mahmud Shah Nawazi (left) and Shams A’din Dadolzai Shah Nawazi (right).
53
54
AMERICAN ANTHROPOLOGIST
[80, 1978
are required to qualify it as a constituent of a segmentary lineage system. One is that brothers and cousins are involved in a more metaphorical than literal sense; what is actually at issue is the relationships between corporate groupings that stand to each other as brothers and cousins through the use of a genealogical idiom. The second elaboration is that none of the alliances are permanent and ongoing, but rat her are contingent upon circumstances, being activated when required but otherwise having no independent existence. The corollary of this is that group membership and identification is defined by circumstance, men giving their all for a particular grouping today and, with circumstances changing and this grouping becoming conceptually irrelevant, acting as members of a quite different group tomorrow. What is clear from the ethnographic evidence, of which the saying quoted above is the merest speck, is that various peoples state that their loyalties and patterns of alliance do indeed correspond to those of the saying, and t o what we have called complementary opposition. This is stated quite vehemently and justified quite subtly; about these statements there is no question. But there is a question, and that is whether these peoples do what they say and are saying what they in fact do. It is this question and its corollaries that are the subject of this essay. The corollaries are two: if complementary opposition is asserted but not acted, what are the actual patterns of loyalty and alliance and group formation, and what factors underlie the actual patterns? And if complementary opposition is present in some cases, under what circumstances is it present and under what circumstances is it not present? To raise this issue is t o follow, at least initially, a well trod and expertly cleared path. Peters, in his masterly paper, “Some Structural Aspects of the Feud among the Camel-Herding Bedouin of Cyrenaica,” has directly attacked this problem, as well as several related theoretical issues. A discussion of Peters’ findings is perhaps the most suitable place t o begin our exploration of the problem of complementary opposition.’ Peters’ argument is directed toward two important strands in social anthropological thought. One is what he calls “lineage theory,” which was developed in Evans-Pritchard’s work on the Nuer (1940) and Fortes’ on the Tallensi (1945), and which posits total societal political systems based upon the balanced opposition of lineage segments related genealogically. The second strand, represented in the work of Gluckman (1959) and Colson (1962), is what might be called “the peace in the feud” theory, which emphasizes the social control aspect of conflict in tribal society, and which sees cross-cutting kinship ties as dampeners of conflict. Peters presents ethnographic evidence from the Bedouin of Cyrenaica which, in his view, throws doubt on the validity of these two theoretical positions. The Bedouin of Cyrenaica themselves present a “lineage theory” model of their society and are able to adduce various bits of evidence to support their view. But Peters says that a systematic review of the evidence indicates quite clearly that the Bedouin act in a fashion contrary to the model: they do not maintain equivalent relationships with groups of structural equivalence; they do not ally with groups structurally close against groups structurally distant; they do not combine in groups of a higher structural level and act corporately. Peters concludes that the lineage model is simply incorrect, an elegant folk ideology that has been mistakenly adopted as an analytic sociological model. The “peace in the feud” theory is incorrect for the Bedouin of Cyrenaica because feud cannot be resolved; feud is unresolvable because it is a relationship of ongoing political conflict over societal resources. The issue in Bedouin feud is not that peace is broken and must be restored, but that groups in conflict over scarce resources are trying to expand or resist expansion. For this reason, cross-cutting ties are the stuff of strategic alliances systematically constructed to further the conflict rather than to dampen it. The Bedouin patterns of alliance reported by Peters are as follows: the basic operative units are tertiary lineage-based groups of several hundred people, each such group being discrete from all others and each fully corporate. The tertiary group has its own wataiz,
Salzman]
COMPLEMENTAR Y OPPOSITION
55
homeland, and its own watering point, ploughland, and pastures. Tertiary groups use marriage ties to form alliances with selected groups genealogically and spatially distant, while maintaining feuding relationships with other groups structurally equivalent to their allies. Thus each has allies among tertiary groups of collateral secondary lineages, and at the same time maintains a feuding relationship with other tertiary groups of the same secondary lineages. More closely related tertiary lineages of the same secondary section do not necessarily act together, even if the matter at hand is one that involves as opponents groups of distant collateral segments. In conflict cases, a tertiary group may be supported in strength by members of genealogically distant groups, while members of certain quite close groups are conspicuously absent. Thus nonpatrilineal ties, such as matrilateral and affinal, are drawn upon, are activated in conflict and other situations, and these links seem to override the patrilineal ones which would dictate different solidarities. Finally, the higher order sections, the secondary, primary and the tribe, do not seem to function corporately either in conception or in action. When a conflict develops or a homicide takes place between tertiary groups of collateral secondary or primary sections, it is regarded as a matter between two tertiary units, not between secondary or primary sections which involves all members of those sections. Men act as members of their tertiary groups, not as members of secondary or primary sections. Likewise, members of secondary and primary sections and tribes do not a m a s and act as one; in any political or military show of force, some tertiary units of a larger section are present in strength and others are absent, and there is often an admixture of allies through other than patrilineal bonds. To sum up, the actual patterns of alliance found among the Bedouin of Cyrenaica are contrary to the genealogically based model that they assert; complementary opposition is a core principle of their model but is not found in their behavior. What is the reason for the actual patterns of alliance “on the ground”? The answer to this, according to Peters, is also based upon what is found “on the ground” in a more literal sense. It is the material interests of the Bedouin, especially in relation to the resources of the environment from which they gain their livelihood, that determine patterns of alliance. One such material interest is in having access to resources such as pasturage and water distant from their home territory, so that localized droughts and other threats t o livestock and livelihood can be avoided through movement to territories of allies. Genealogically close tertiary groups would be unlikely to help in such cases, for being spatially adjacent, they are likely to be under similar environmental pressures. A second material interest is in being able to expand one’s territory as human and animal population increases or as local resources decline. Such expansion would have to be at the expense of neighbors, who are likely to be collateral tertiary groups of one’s secondary section. Thus both of these material interests, in temporary access and permanent expansion, direct one t o look t o spatially and therefore genealogically distant tertiary groups for alliance and direct one away from spatially and genealogically close tertiary groups. In this way, material interests contradict the injunctions of lineage theory and segmentary opposition, and although the Bedouin repeat the injunctions, they form their alliances and otherwise act in accordance with their material interests. But why do the Bedouin repeat the injunctions of lineage theory, why do they hold to, in Peters’ words, “the simple ossified structure of their genealogical relationships” (1967:275)? Peters’ explanation (1967:270) is that lineage theory “. . . is a kind of ideology which enables [the Bedouin], without making absurd demands on their credulity, to understand their field of social relationships, and to give particular relationships their raisorz dPtre.” Are the Bedouin, then, so unaware that many of their acts and alliances are discordant with their ideology? Of course not, but they see these various acts as exceptions based upon unusual and contingent circumstances, and they do not see them as evidence that their lineage model corresponds very little with behavioral reality. This is where the Bedouin are incorrect.
56
AMERICAN ANTHROPOLOGIST
[80, 1978
“What they fail to appreciate is that these ‘contingencies’ are ecologically, economically, demographically, and politically essential” (Peters 1967:275),and thus that their “irregular” alliances and adjustments are necessarily the core reality of their social arrangements. Bedouin lineage theory must therefore be seen as an institutionalized cognitive crutch with little relation to social reality. Although Peters’ account of the Bedouin of Cyrenaica is not documented with case material or substantiated by quantitative analysis, I do not think that it would be unduly credulous of us t o accept his description at face value, in general outline at least. Nonetheless, there are a number of points about the workings of the system as Peters describes it which might be questioned. The first is about territorial expansion. Competition for natural resources is said to exist between spatially adjacent collateral lineages, and this undermines solidarity between collateral tertiary groups. But are neighbors always collateral tertiary groups? Some neighbors of some groups must be from collateral secondary or primary sections. One would like to know what percentage of tertiary groups have neighbors who are not collateral tertiary groups. Those tertiary groups with different groups of neighbors of varying structural distance would be able t o choose to move against genealogically more distant groups and to minimize competition with genealogically close groups. Furthermore, tertiary groups surrounded solely by collateral tertiary groups could press their collaterals who do have more genealogically distant neighbors to expand at the latter’s expense and thus make room for the closely packed collaterals. Military assistance could be offered by the hemmed-in collaterals, and the genealogically based alliance could result in more territory and resources for some or all of the members of the secondary section. Now perhaps such alliances do not take place; Peters reports no such pattern. But there is nothing in Peters’ account that would sufficiently explain the absence of such a pattern, certainly not the assertion about propinquity of collateral tertiary segments, which must impossibly oversimplify the actual pattern of territorial distribution of lineage segments. The second point is also about competition and expansion. Peters argues that in spite of resource competition between neighboring collateral groups, the competition cannot be allowed t o develop into open conflict. Stultified in the development of their hostilities by virtue of their territorial proximity and the sentiment associated with putatively close agnatic connexion, A and B are able to express hostilities indirectly through their [spatially and genealogically distant] linked groups X and Y respectively [1967:277]. Now one can see the symbolic and psychological benefits of defeating proxies, but how does this assist in expanding one’s control over resources? As Peters says, “The capture of natural resources for use by a tertiary group can only be achieved by an enlargement of its homeland, and this is only possible by expanding into a neighbour’s territory” (1967:274). But if neighbors don’t fight, how does this expansion happen? What is the mechanism? The third point is about patterns of alliance, about complementary opposition, or the lack of it, among the Bedouin of Cyrenaica. Peters’ general conclusion is that “groups do not come together in their respective genealogical orders” (1967:278).He observed a number of disputes in which members of different higher order sections were involved: “In every case that I recorded the agnatically closest groups of the main participants were absent, while members of genealogically more remote groups were present in strength” (1967:278).Now this I regard as strong evidence in favor of Peters’ general position. But I would also like to know if genealogical ties and complementary opposition did not play some kind of restraining role in the commitments of the active and inactive parties. Were the options of some groups limited, and were some groups in effect neutralized by genealogical ties? If some groups did not do what they “should” have done, did they also avoid what they “should not” have done? Might a tertiary group feel free to withhold support from a rival
Salzman]
COMPLEMEN T AR Y 0PPOSI TI 0N
57
tertiary group, but at the same time feel obliged t o abstain from supporting the genealogically more distant opponents of its rivals? If this were the case, I think we would want to know more about it than Peters tells us. It would not be evidence that the lineage model is correct, but it would indicate that the genealogical ideology has more efficacy than Peters seems to suggest. And it would make Peters’ statements about the inability of genealogically close neighbors to engage in hostilities somewhat less anomalous. There is one further point that I would question, and this is less a factual question than one of interpretation. I am not satisfied with Peters’ explanation for the strength of lineage ideology among the Bedouin of Cyrenaica. Why do they so vehemently hold such an incorrect ideology? In what sense can they “understand their field of social relationships” (1967:270) through this misleading model? One might understand the presence of a false model in a hierarchical society, where the interests of the elite were furthered by it, and where the elite actively propagated it. Or one might not be surprised at finding a “false model” about the genesis of illness or the structure of heavenly bodies, where systematic empirical inquiry is beyond the technical resources of the society. But would we really expect a fully incorrect “cognitive crutch” about conflict and alliance in an egalitarian society, where men spend a good deal of their time arguing, planning, plotting, arranging, manipulating, and managing these very things? It is true that the Bedouin are not sociologists, but there must be more to their incorrect model of their society than that; there must be more to explain the genealogical model than the fact that it is good t o think. One element of an explanation might lie in the restraining function of genealogical ties suggested above. The dampening of conflict that Peters says does not come from cross-cutting ties may come from genealogical ties which are contrary to material interests. If this were so, then the holding of the genealogical model would, in spite of its sociological incorrectness, be an independent and efficacious motivating force in the more complex alliance and conflict patterns of the Bedouin. In other words, the genealogical model is held at least partly because it contributes to a more peaceful and satisfactory social life. Now this argument would depend upon evidence about the constraints imposed by the genealogical model, and no such evidence is presently available. It does deserve examination, if only as a replacement for the weak “cognitive crutch” argument. But I carry no special brief for it. Indeed, I shall later return to this subject with an alternative explanation. Even with these various queries and qualifications, I think we have to agree with Peters that the Bedouin do not conform in most respects t o the lineage model, that complementary opposition is not a dominant feature of their alliance system, and that knowledge of many nongenealogical aspects of Bedouin life is required for even a minimal understanding of conflict and alliance patterns. Are these findings about the Bedouin of Cyrenaica applicable to all Bedouin, and to all peoples who claim to have, or are described as having segmentary lineage systems? Is it universally the case that the lineage model is a pure ideological fabrication with no application in behavioral reality? In the next section of this paper I will address these questions to several cases from the ethnographic literature. The Yomut Turkmen of northeastern Iran, as described by Irons (1975),are in gross outline similar to the Bedouin of Cyrenaica in being nomadic, in herding livestock and engaging in some cultivation, and in having a genealogically based ideology, an acephalous segmentary political system, and territories associated with groupings at lower and higher orders of segmentation. Yomut Turkmen groups, like those of the Bedouin of Cyrenaica, are allied with other Yomut groups spatially distant against neighboring Yomut groups. This pattern differs from that of the Bedouin in two important respects. First, the units involved are of a higher order of segmentation. The Turkmen equivalent of the territorially based tertiary group of the Bedouin is the oba with a population of several hundred. These oba are grouped spatially, politically, and to a large degree genealogically, into larger units called il, which is the largest identifiable residence group, consisting of from five to thirty oba. The
58
AMERICAN ANTHROPOLOGIST
[80,1978
term il means, in addition to a group of this size, a relationship of peace. It is with this larger group, the ill that we will be primarily concerned. The second difference is in the relationship between the genealogical framework and the alliances between distant groups against near neighbors. Among the Turkmen, the territorially based il are combined into larger groupings also called il, which can be thought of as nonterritorial confederacies. These larger groupings are nonterritorial because the constituent units are not contiguous, but rather interspersed with the constituent units of the other confederacy, so that each small territorial il has as neighbors on one or two sides members of an opposing confederacy. Most important here is that the oba, the territorial il, and the confederacy il are defined largely in terms of genealogy, so that the near neighbors of territorial il are distant genealogically, and more distant territorial il are close genealogically. This means that one is allied with distant groups against neighboring groups while at the same time one is allied with genealogically close groups and in opposition to genealogically distant groups. There is at the level of the Yomut territorial il no contradiction between spatial distance and genealogical closeness and no contradiction between conflict with neighbors and solidarity with closely related kin groups. The Yomut have institutionalized, albeit at a slightly higher level of segmentation, Peters’ insight that conflict over resources is likely to arise between neighbors, but they have done it by alternating close kin groups with distant ones, so that genealogical solidarity can be maintained in spite of conflict between neighboring groups. It seems likely that manipulation of upper levels of the genealogy played a part in the development of this pattern, for the neat alteration of constituent units of the confederacies is otherwise hard to explain (Irons, personal communication). If so, it is that symbolic manipulation which allows the Yomut to act according to a genealogical model and to the injunctions of complementary opposition. The Yomut do not have a pure segmentary lineage system, for although territorial units have core lineages, strangers of other lineages and in a few cases stranger lineages are also present. These strangers act along with the core lineages of the territories in which they are resident, and after several generations are often incorporated into the core lineage. As Irons says (1975:58): [the structure can be best] described as the segmentary political system of the Yomut. This consists of a series of named territorial groups and subgroups that is very much like a segmentary lineage system in its political aspects. At each level of segmentation groups can be mobilized in opposition to one another or united into a single coalition against a group of the next higher level of segmentation. The system differs from a segmentary lineage system, however, because it is only partially based on a genealogy [as described above]. Nonetheless, “there is a tendency for the two [lineages and territorial groups] to correspond approximately in composition” (1975:49). Within this system, the Yomut conform relatively closely to the dictates of genealogical/territorialsolidarity and complementary opposition. Competition for resources between oba is virtually nonexistent, and individual conflicts are infrequent and the subject of strenuous peacemaking efforts. Conflict over resources between neighboring territorial il of different confederacies is relatively frequent and intense. And in such conflicts, the oba of each territorial il unite for the conflict. If hostility between two tribes became intense enough, for instance in a dispute over territory, each tribe would organize a military expedition known as an aq oyli. The aq oyli was constituted as follows: each oba would send one-fifth of their adult males as a fighting contingent and one-fifth of their yurts to house them, while food would be supplied by the people left behind. The total group would elect a leader, known as a beg, and under his direction they would form a highly mobile fighting force [Irons 1975:65]. Even in the present postpacification context, groups unite on the same lines for purposes of nonviolent political action. This is not to say that all Yomut always rush to support their
Salzman]
COMPLEMENTARY OPPOSITION
59
kinsmen or that all groups are enthusiastic to support close kin groups in every dispute. Some individuals and groups sometimes attempt t o stay out of a dispute, especially if they think it is not important or has little payoff for them. But the demands of reciprocity, the need for allies in the future, usually result in Yomut actively supporting their closer kinsmen, if sometimes grudgingly. And no cases of Yomut groups actively supporting distant kin groups against closer ones have been recorded. “Turkmen social structure allows very little room for choice in matters of political alliance” (Irons 1975:114). And individuals who, for economic or other reasons, ally contrary to the dictates of genealogical/territorial loyalties usually find themselves in severe trouble with their fellow kinsmen (Irons, personal communication). The spatial checkerboard pattern of alliances seen among the Yomut Turkmen and the Bedouin of Cyrenaica is also found among the Swat Pathan (Barth 1959), a people who in most respects are unlike the Turkmen and Bedouin, having a highly stratified and hierarchical political system and an adaptation based upon intensive irrigation agriculture. The checkerboard pattern of alliance results from the importance of territorial resources and the conflict between neighboring groups over those resources. Since in this society, neighbors are often close kinsmen, the Pathan khan is often allied with distant kinsmen against close kinsmen. But there is no ideology of genealogical solidarity in this society; the genealogy functions only as a pedigree to justify the elite position of the khans in relation to the rest of the population. Complementary opposition exists here neither in thought nor action. The Somali of the northern part of the Horn of Africa, described by Lewis (1961), are nomadic camel herders with a genealogically based acephalous political system. Unlike the Bedouin of Cyrenaica and the Yomut Turkmen, the pastoral Somali do not have stable territories at levels of segmentation lower than the clan, which is composed of up to 100,000 individuals. Somali lineages of deeper and of shallower degrees of depth below the clan overlap spatially at any given time and vary in their distribution from time to time, season to season, and year to year. No rights of ownership are held in territory, control being based upon effective occupation, the ability to defend the territory. The Somali political system is based upon the solidarity of patrilineally related groups against patrilineally more distant groups. Although agnation, tol, provides the framework of the system, it is modified by the use of two other principles, bah, uterine alliance, and heer, contract. Uterine alliance and fission, based upon the recognition of mothers rather than fathers as apical ancestors, is a common supplement to considerations of agnation. “Uterine ties are remembered or invoked t o strike alliances across agnatic cleavages to redress the balance of irregular growth and uneven proliferation’’ (Lewis 1961:159). It is thus a mechanism used to counter demographic processes which render agnatically based lineages unequal and which upset the balanced opposition of segments upon which the political system must be based, Especially weak lineages not able t o use uterine alliance will take a dependent status through a gaashaanbuur alliance outside of kinship; such arrangements are sometimes necessary but regarded as without honor. Heer, contract, is used to crystalize the political obligations of agnation. “Political and jural responsibility is defined by contract within agnation” (Lewis 1961: 167). Lineages of various depths use contracts among their members to specify “. . . the delicts and punishments which they wish to recognize amongst themselves and establish the principles by which they will act externally as a collective political unit” (Lewis 1961:176). The general obligation of solidarity is inherent in agnation; the contract specifies the operational details within the kinship framework. “In the relations between lineages which regularly meet in their grazing movements there is a perpetual cycle of aggression and conciliation” (Lewis 1961:246). Due t o the absence of stable territories, there are no spatially close neighbors or distant neighbors. There are only those tribesmen who meet at a given time at a given point. They may be close kin or distant
60
AMERICAN A N T H R OPOLOGIST
[ S O , 1978
kin, bound by common contract or unbound. When pasture and water are scarce, as they regularly are in Somaliland, distant kin may not be very welcome to the limited resources that one had planned to use. Where there are contributory factors such as competition for the same resources, what begins as a dispute between individuals of different clans and spreads in the dia-paying [ blood-money paying] groups immediately concerned, eventually culminates in generalized clan enmity. . . As hostility increases in range involving an ever widening circle of kinsmen until the whole clan is implicated, the members of each of the opposed groups increasingly view each other as undifferentiated wholes. . . . Enmity spreads beyond the centres of conflict colouring the relations between all members of the opposed clans wherever they are. . . . The sense of clanship is so strong that men naturally identify their interests with those of their kinsmen who are actively engaged in war even when they are themselves [as government officials] concerned to maintain the peace [Lewis 1961:253-2541.
.
There is not only a “natural” sense of loyalty and identification involved, but a sense of self-preservation, for “the pastoralist’s general security rests upon the fighting strength of his dia-paying group and of the lineages and clan to which he belongs” (1961:255). The material interests of each Somali rests with his agnates. There are no contradictions between propinquity and agnation as there seem to be for the Bedouin of Cyrenaica. The Somali have no neighbors; they have only close and distant agnates and resources over which they are or are not able to establish effective control. Before going on to the final ethnographic example, let us pause here and reflect on the significance of the cases already presented for the problem of complementary opposition. Peters, in addition to presenting ethnographic evidence that complementary opposition is not found among the Bedouin of Cyrenaica, constructs a number of general arguments against the possibility that “lineage theory” could be realized in behavioral reality. Two of these arguments have to do with balanced opposition. The first one is that there are usually more than two segments at a given level, and this means that balanced opposition is impossible because “combinations can occur which are not comprehended by the theory” (Peters 1967:271). This does not seem to me an objection to lineage theory, but only a statement that patterns of social behavior other than those prescribed by a lineage ideology can exist. This is undeniable; indeed, if the lineage ideology covered every possible pattern then it would not be a substantive ideology at all. If two lineages descended from two brothers combine against a lineage descended from a third brother, then the canons of lineage ideology have been violated. If, on the other hand, two lineages descended from two brothers are in conflict, and the lineages descended from other brothers either stay out of the conflict altogether or actively engage in peacemaking, because they are structurally equidistant from the conflicting parties, then the lineage ideology is being followed. Thus whether lineage ideology is being followed when there are three or more collateral lineages at a given level is an empirical question. That it is more difficult to violate the ideology when there are only two collateral lineages hardly makes this condition a prerequisite for behavior based upon lineage ideology. The second argument about balanced opposition is more serious, for it raises the well recognized point that “all segments should be fairly equal in numbers of people and economic resources, if the theory of lineage were t o have substance in reality” (Peters 1967:271), and that demographic variations result in other variations between lineages. Peters states (1967:271) that “with marked preponderance in numbers and superiority of resources, some groups have grown into dominance, and the evidence points to the existence of this dominance in the past also.” Is balanced opposition then impossible to obtain, or even approximate, in a functioning lineage system? The Turkmen and Somali examples described above suggest that there are mechanisms for redressing demographic imbalances in order to maintain some degree of balanced opposition among lineages. The Yomut Turkmen
Salzman]
COMPLEMENTAR Y OPPOSITION
61
add individuals and groups to core lineages through residential contracts in building their territorial political segments. The Somali use uterine ties for alliance and for fission within their agnatic framework and occasionally use dependency contracts outside of the descent framework. This evidence thus indicates that some significant degree of balanced opposition can be maintained and that considerable trouble is taken in some groups to see that it is so maintained. But here we run into a conceptual problem, for the mechanisms used by the Turkmen and the Somali to counter demographic imbalance are not in fact lineage mechanisms but are rather independent and extraneous principles added to the lineage model. Must we for this reason conclude that these cases are irrelevant to the discussion, and that they do not count as segmentary lineage systems? Certainly it is true that neither the Yomut Turkmen nor the Somali have pure lineage ideologies. They do not assert that they follow agnatic principles exclusively, nor do they deny that other principles are used as supplements. Residence is recognized by the Turkmen as an element additional to descent in political formations. Uterine ties and contract are recognized by the Somali as additional principles which are used within the agnatic framework. These groups do not seem to have incorrect models of their systems. But the cost of having realistic models is that they are not pure lineage models, they are lineage-plus models. In societies with these lineage-plus models, lineage solidarity, balanced opposition, and complementary opposition seem to exist in practice. Perhaps we would wish to use this evidence to conclude that pure segmentary lineage systems d o not exist in behavioral reality and thus to agree with Peters. But whether or not we feel that this is an important point to establish, we must not lose sight of the fact that segment solidarity, balanced opposition, and complementary opposition do seem to be present among a t least some groups with lineage-plus ideologies. And this goes against the expectations that arise from examination of Peters’ material on the Bedouin of Cyrenaica. Peters has two further general objections to lineage theory. One is that due to the lack of institutionalized leadership, “it is not possible for all the sections of, say, a secondary section to come together; there is no one to command the action” (1967:271). The Yomut Turkmen, as described above, elect a temporary official, the beg, to command the action. The Somali meet in councils of male adults and are led by informal leaders, the elders. The fact that there are no institutionalized permanent leadership roles in acephalous societies does not mean that informal leadership or temporary leadership roles are necessarily absent. The evidence from the Turkmen and the Somali does not suggest the general validity of Peters’ observation on the Bedouin of Cyrenaica, “that there are men who wield power over groups of varying populations, and that the extent of their domains cannot be understood by reference to orders in the lineage” (Peters 1967:271). Nor does the evidence from these other groups suggest that a lack of leadership keeps higher order segments from acting in unison. Peters’ final general objection to lineage theory is perhaps the most subtle. Looking especially at marriage and affinal and matrilateral bonds, he argues that each person has a complex bundle of roles and that these all have an impact on behavior, so that “jural rules rarely appear in their ‘pure’ form in behavior” (1967:272). Precedence can only be given to specific forms of behavior in relation to situations. There is no such thing as a general primacy with regard to any form. The concept of agnation as the centre of all political problems, disturbed or disrupted by, say, matrilaterality, is as inadequate as saying that the nucleus which pulls all other components into position is matrilaterality with agnation as a disturbing element. In reality there are numerous components present in all situations, which should not be lumped together under such categorizing terms as agnation and matrilaterality, but which combine as separate components in a kaleidoscopic range of clusters [Peters 1967:272]. This is a complex argument and requires several comments. First, in focusing upon political alliance and conflict, we are concerned with a particular kind of situation and not asserting
62
AMERICAN ANTHROPOLOGIST
[SO, 1978
“the general primacy” of a principle in a society or in all social situations. Second, such an a priori assertion as “all roles are present in each situation and affect behavior” (Peters 1967:272) short circuits what should be a matter of empirical investigation. Surely the question is which elements are present in what situations, and to what degree. It is quite conceivable that matrilaterality is more important in some situations than in others, and in some societies more than in others, Peters’ position is not an effective argument against the possibility of lineage-based patterns in political behavior, but it is a fair warning that various factors can enter into political behavior, and we had better know what they are before we say that political behavior is lineage based. Third, Peters seems to suggest that all elements present must be seen as having equal weight, so that we will not fall into the trap of defining one element as a “disturbing” one. I would argue that we can be more subtle in our analysis without defining elements as intrusive. For example, there is a difference in the weight of affinal ties between situations in which a man will not support his agnates against his affines and in which a man will actively support his affines against his agnates. Likewise, we may conclude in some cases that lineage solidarity was indeed the dominant factor. Among the Somali, for example, affinal bonds seem to count for little in situations of political conflict. Fourth, categorizing terms such as agnation can indeed be more trouble than help, and one must be careful in using them, especially avoiding ideal typical constructions in which a great deal of baggage is defined in and then assumed as necessary. Likewise, one must avoid simplistic characterizations that seize upon one feature and make it the core of the system being studied. However, we cannot avoid categories if we are to think and communicate, and if we are to rise above particular cases to general understandings. I think that we will need more guidance than Peters’ statement that we should think in terms of “components in a kaleidoscopic range of clusters.” These general arguments of Peters seem to me t o be effective only if discussion is limited to fully pure lineage systems. The lineage-plus systems described above are perhaps one kind of evidence that pure lineage systems are impossible. However, these lineage-plus systems do, as mentioned earlier, have segment solidarity, balanced opposition, and complementary opposition, which the Bedouin of Cyrenaica do not have according to Peters. Why should these patterns be present in some cases and not others? The reasons for the presence of these patterns among the Somali and their absence among the Bedouin of Cyrenaica seem to turn on territorial stability, the presence of which in the Cyrenaican case provides a basis for political relations inconsistent with the lineage model, and the absence of which among the Somali provides a basis for political relations through the lineage model. The Somali, moving about in irregular patterns as a response to erratic resources, cannot establish political relations on the basis of propinquity. The lineage model relates all Somalis to all others no matter what may be their distribution on the ground. Whoever a Somali meets can be placed as politically close or distant and consequently as welcome or unwelcome. Furthermore, the lineage system provides a stable framework of group political support while allowing group members to organize in small flexible local groupings for the purpose of day-to-day adaptational activities and to change membership and movement patterns as adaptational and demographic processes require. The Bedouin of Cyrenaica, according to Peters, are territorially stable and, in competition for natural resources, are most likely to come into conflict with their neighbors, who are also close collateral groups. The material interests of these Bedouin thus lie in opposing close kin and allying with more distant kin. This is also the pattern among the Swat Pathan, who are territorially stable and for whom control over territory is the primary material prize of political competition. The Turkmen have managed to combine both stable territoriality and a segmentary political system by arranging close and distant groups in a checkerboard pattern, thus allowing competition between spatially close but genealogically distant groups and alliance with genealogically close but spatially distant groups. Territorial interests and segmentary solidarity are in harmony among the Yomut Turkmen.
Salzman]
COMPLEMENTA R Y 0PPOSI T I 0N
63
The puzzle of lineage ideology still, however, remains. Why do the Bedouin of Cyrenaica continue to assert that they follow a segmentary lineage model, when in fact they do not? Why, for that matter, do the Turkmen bother to keep their genealogy adjusted to their territorial pattern? If territorial interests are the core of political competition among these groups, why is this not reflected in their ideology, as it is among the Swat Pathan and numerous other territorially based groups, and why is not the misleading lineage ideology scrapped? One could argue, of course, that the lineage ideology is a survival from times past, and that in due course it will be scrapped. But this kind of argument is especially suspect in an area such as politics, which is a matter of daily discussion, argument, and conscious reflection. No, I will argue that the lineage model is maintained because it provides a framework, not for common sense understanding, as Peters would have it, but for social mobilization in circumstances which remove the territorial commitment from consideration. Peters argues that the Bedouin of Cyrenaica are incorrect in viewing their deviations from the lineage model as exceptions which result from contingencies but which do not affect the rules. “What they fail to appreciate is that these ‘contingencies’ are ecologically, demographically, and politically essential” (1967:275). I would respond with the suggestion that these “contingencies” are essential only as present circumstances obtain, especially present circumstances in relation to territories. The “contingencies” may be “essential” only in the short run, and they may disappear in the long run. If they disappear as a result of changed circumstances, they leave the lineage model as a basis for mobilization. Peters says, “I see no reason to assume that social relations as I observed them are present today or that they were the same at any time in the past either” (1967:280). Much of the history of tribal North Africa consists of large-scale territorial displacement and population movement. In such circumstances, a tribal population would do well t o have a way of ordering itself independent of territorial distribution. For this thelineage model is, if not ideally, at least very well suited. The Bedouin of Cyrenaica can be seen as maintaining their lineage ideology as a social structure in reserve, as a nonterritorial system of organization which could be used in the case of spatial dislocation. Nomads, of course, are used to moving about and moving their capital resources about. Migration of a nomadic population to a near or distant area for political or economic reasons is often a preferred option rather than an imposed disaster, as it might be for a sedentary population. That a nomadic population would wish to keep its options open, including organizational concomitants of nonterritorially based adaptations, seems something less than farfetched.2 But if this argument about the lineage model as a social structure in reserve continues to sound farfetched, let us return to the Turkmen case, for here the historical evidence bears out such a formulation. The Yomut maintained a nomadism far beyond their requirements for adaptation on the Gorgon plain. This nomadism was not primarily an adaptation to the physiobiotic environment, but rather to the sociopolitical environment. The Yomut were on hostile terms with the Persian government; they raided Persian peasant villages and Persian caravans and wished to stay independent of the Persian tax officials and army recruiters. The Persian government sometimes sent armies against the Yomut, and the Yomut would sometimes make a strategic retreat across the Kara Kum desert and take refuge with the Khiva Yomut. There are records of this having taken place from the 16th to the 20th centuries (Irons 1975: 72-73). Being displaced from their stable territories and having to rely upon nonterritorial bases of organization was a regular if not frequent, and expected if not desired, part of Yomut Turkmen life. Is it any wonder that their genealogical model for social organization was kept in good repair? The final ethnographic example is the Shah Nawazai tribe of Iranian Baluchistan upon which I have done r e ~ e a r c h .This ~ case will support some of the arguments made earlier but is included primarily to explore more deeply the conceptual and methodological problems in the study of segmentary political systems and complementary opposition in particular. One question raised earlier in this paper was whether the Turkmen, the Somali, and similar
64
AMERICAN A N T H R 0POL OGIS T
[80,1978
groups with strong lineage systems supplemented by additional elements, such as uterine ties and contract, can be legitimately considered counterexamples to Peters’ arguments about lineage theory. Part of the problem is that Peters does not distinguish between the Bedouin lineage ideology, Bedouin lineage ideology masquerading as a sociological theory, and a genuine sociological model of segmentary political systems. Peters argues that we are taken in by a Bedouin ideology which is an incorrect representation of reality but neglects to consider that some ideologies might be fairly accurate; those of the Turkmen and the Somali do seem to be fair representations of reality, partly because they recognize nonagnatic elements and partly because the ideologies fit or have been adjusted to the exigencies of social and ecological reality. Peters says that “once contingencies are permitted to enter in, the lineage model ceases to be of use” (1967:271). This might be a fair comment on an uncritical adoption of pure lineage ideology, but it is hardly a useful guide for constructing a sociological model; after all, contingencies are not random and beyond understanding, but patterned in relation to certain circumstances, and as such can be incorporated into an analytic model-not, perhaps, a segmentary lineage model, for pure lineage systems do not seem to exist, but a segmentary political system model. Questions about the existence in behavior of segmentary solidarity, collective action and complementary opposition could be most fruitfully dealt with in relation to a segmentary political system model rather than a lineage system model. Tentative answers, based upon the case material presented here, would be that these mechanisms are present among some groups, that they operate with some degree of effectiveness, and that to some substantial extent they are at the base of political actions and patterns (cf. Gellner 1973:3-6). The Shah Nawazi Baluch are sheep- and goat-herding nomads who also engage in other forms of extraction, such as cultivation, and in the past raiding, and currently several types of market-related labor. They have a curiously mixed political structure, rather like a segmentary lineage system with a chief on the top. The office of chief is very limited but adds a measure of central direction in some spheres, especially external relations. Internally, many sectors, notably the economy and social control, especially blood responsibility, are handled in a decentralized fashion. There is a strong lineage ideology, although other kinds of ties are acknowledged as being extremely important. The tribesmen say explicitly that the importance of the different kinds of ties depends upon the circumstances; it is said that in political matters agnation is everything and affinal and matrilateral ties are nothing. But what happens in practice? Several years ago, while the tribesmen were resident at their lowland date groves during the summer harvest period, Mahmud Karim, one of the Dadolzai lineage, discovered that a palm log he had cut and prepared for a hut rafter had been carried off by a member of the nearby Kamil Hanzai lineage. Mahmud Karim was furious and went about gathering support from his fellows for an excursion to retrieve the log. One person he went t o see was Ja’far, the most politically senior man in his lineage, who gave his blessing to the excursion with the proviso that the log was the object and that a fight should be avoided unless the retrieval was opposed. One point worth noting is that Mahmud Karim and Ja’far are not close kin but are rather distant as lineage mates go. Ja’far was consulted as a senior man in the Dadolzai; the slight was thought of as being to the Dadolzai, the collateral lineage to the Kamil Hanzai. That is, structurally equivalent lineages were being activated. A party of 12 tribesmen armed with sticks, knives, axes, and stones went off for the log, which was retrieved without incident. Among the 12 men all of the sublineages of the Dadolzai were represented. But 4 of the 12 were not Dadolzai! Two were Gamshadzai, from an entirely different tribe; they were long-time associates of the Dadolzai and were staying with the Dadolzai at the date groves. Two were affines, sons-in-law of Ja’far, and so sons-in-law of the Dadolzai, but not close kinsmen of Mahmud Karim. One of these sons-in-law was related matrilaterally to the Kamil Hanzai, but he resided with the Dadolzai. This was true of Ja’far himself, who said
Salzman]
COMPLEMEN TA R Y 0PPOSI TI ON
65
that agnates must stand together, even if against matrikin. No members of the Mogolzai lineage, which was spatially contiguous with the Dadolzai but structurally equidistant from the Dadolzai and Kamil Hanzai, joined the Dadolzai “war party.” Now how are we to think of this group of men? Was it truly a Dadolzai group; can this count as lineage collective action? What do we make of the fact that resident affines and resident outsiders joined but contiguous co-residents did not? Are numerical considerations, the eight and four split, important in evaluating the significance of this grouping? The event just described was quickly followed by another involving the same groups. A member of the Kamil Hanzai got into a fight with a member of the Rahmatzai, one of the collateral sections of the tribe, structurally equivalent to the Soherabzai of which the Kamil Hanzai, Dadolzai, Mogolzai, and many others are constituent lineages, over a date-eating camel. After this fight, in which the Rahmatzai was worsted, some other Rahmatzai roughed up two elders of the Kamil Hanzai whom they happened to come across. This sent the Kamil Hanzai into paroxysms of indignation. They came around to the Dadolzai and other Soherabzai lineages demanding support, and the other lineages agreed, for after all, they were “one,” they were all Soherabzai. News came that the Rahmatzai were massing t o meet the Soherabzai in battle, and the Soherabzai rallied together, sending a “war party” of 120 or so lightly armed tribesmen (and one anthropologist) to meet the Rahmatzai, who did not show up. There were some 30 Kamil Hanzai in this party, the remaining 90 being from other Soherabzai lineages in the vicinity. During this period, a Rahmatzai son-in-law residing with the Dadolzai fled back to the highlands out of fear. He had been threatened by the Kamil Hanzai, and although the Dadolzai said that he was under their protection, he was plainly uncomfortable residing with the Soherabzai during this time. Several points can be made about these incidents. One is that the events were not considered solely the business of the individual smaller lineages initially involved. Second, conceptual complementary opposition was in operation from the very beginning; the opponents were never referred to as anything but Rahmatzai. And when “our” side was referred to, it was always “we Soherabzai.” Third, the Soherabzai in this rather large section of the date grove area did come together and act as a unified action group. Taking these three points together, it does not seem unreasonable to suggest that something rather like segment solidarity, complementary opposition, and unified action was taking place. Admittedly, not much can be concluded from this highly truncated account of a single set of incidents, but would it be an exaggeration to say that there are things happening here that we would not be led to expect among the Bedouin of Cyrenaica according to Peters’ account? Of course, the Shah Nawazi Baluch are different from the Bedouin of Cyrenaica in crucial respects. The one most relevant here is their lack of territorial stability, not withstanding the date groves. During nine months of the year which they spend on the highland plateau herding their animals, the Shah Nawazi lineages are highly mobile, limiting themselves to no special portions of the tribal territory. This “open pasture” policy of the tribe is related to the variability and unpredictability of the meager rainfall and the consequent pasturage. Thus the Shah Nawazi lineages are mixed and remixed throughout the year and from year to year, in a fashion similar to the Somali, and they likewise order their intergroup relations through a genealogical model. In the light of these data, I think it would be desirable to reformulate our questions about complementary opposition and segmentary lineage systems. It does not seem to me fruitful to argue, as Peters does, that unless there is an ongoing one-to-one correspondence between ideological statements, such as about complementary opposition, and behavior patterns, such as alliances and support activity, the ideology should be disregarded by the observer lest it mislead him about the nature of social reality. Such an approach does not do justice to the complexity and subtlety of the relations between ideas and behavior. Peters rejects the possibility of lineage-based complementary opposition existing in pure
66
AMERICAN ANTHROPOLOGIST
[80, 1978
form on the behavioral level and, by disregarding the role of ideas, argues that behavior patterns reflect “numerous components. . . in a kaleidoscopic range of clusters” (1967:272). Such a position leads to approaching complex social phenomena in terms of two kinds of questions. First, there are those which are aimed a t ascertaining whether a behavioral pattern, in this case, complementary opposition, is present or absent: How many affines spoil a complementary opposition? What percentage of agnates have to turn up for action in order that it be considered unified action? Second, there are those questions which are aimed at ascertaining all of the components that might play a part in a particular instance, in this case, of alliance and support: Do affinal ties come into play in alliances? Are matrikin found in support groups? The problem with such an approach is that it swings between extremes, from a demand for a “pure” pattern to an assertion that all elements can be considered of equal importance. Such an approach does not lead to fruitful analysis or generate informative explanations. In contrast, I would argue that it is more useful to ask to what extent a particular principle or pattern is present, and to what degree it is modified by this or that factor or circumstance. And this entails developing and refining ways of investigating and expressing degrees of presence and of influence. This involves at least two levels: that of description, the observable patterns of behavior, and that of inferring influence and constraint, the underlying factors that give rise to the behavior. Through such analysis, it should be possible to establish the extent to which a principle of organization is exhibited on the behavioral level, and the extent to which the observed pattern results from an ideological principle (or principles) held by the actors. In examining the extent to which complementary opposition is present in a particular society, it is necessary to ask, initially, as Peters does, what the ideal and the actual patterns of alliance are. But it is necessary to go beyond this, especially if the constraining effects of ideological principles are to be ascertained. The actual pattern of behavior found must not only be compared with the pattern asserted by the ideological norm, but it must be placed in the range of possible behavior patterns. This is because it may be quite as significant that certain patterns do not emerge as it is that others do emerge. This is easy to illustrate in regard to the question of alliance patterns: Group A does not actively support genealogically close group B in the latter’s dispute with genealogically distant group X. Group A thus does not conform to the ideological injunction that it should support genealogically close groups against distant ones. However, it is equally noteworthy that it does not ally with group X against group B. One possible explanation of the actual pattern is that the material interests of group A and group B are in conflict, and this accounts for the lack of support. But this does not explain why group A does not actively support group X against group B. An alternative explanation is that the interests of A and B are in conflict, and that on the basis of interests alone A would support X against B. But A’s ideological commitment to genealogical solidarity and complementary opposition militate against such an alliance. The actual pattern, then, could result from an interplay between material factors and ideology, or, as might be the case, between short-term material considerations and long-term material considerations represented by the ideology. In the case of most social patterns, there are many more than three alternatives, and the number of influencing factors more than two. In the case of alliances and support, one can envisage a scale of many degrees of support/nonsupport/opposition,rat her similar to social distance scales and, like social distance scales, based upon the judgments and evaluations of the members of the culture being studied. In political support, the scale might range from full material and group support, to partial material and group support, to material support alone, to verbal support, to apologetic lack of support, to defiant lack of support, to verbal opposition, and so on. The particular position that a group or set of individuals might take on a particular occasion could, even if the position were not the one demanded by the
Salzman]
COMPLEMENTARY OPPOSITION
67
ideology, be significantly influenced by the ideological factor among the others at work. This is a point implicitly recognized by Peters (1967:277) in his explanation of the absence of direct conflict between collateral groups (quoted above). The conclusion that must be drawn is that if ideology is not the sole determining force in actual patterns of behavior, it may nonetheless play an important guiding and constraining role. Thus ideological formulations, such as those asserting complementary opposition as a guiding principle, can be most fruitfully seen not as ideational screens obscuring the behavioral reality but rather as active factors influencing and constraining the choices that generate behavior patterns. There is another way in which the pure pattern and equal element approach inhibits an understanding of the significance of ideological principles. By discounting the role of such principles, this approach neglects what is perhaps the most important question: Under what circumstance does a particular principle arise, and under what circumstances is it more constraining and under what circumstances less constraining? In the case of complementary opposition, we have seen that its significance and impact on the behavioral level are related t o territorial stability and instability. Thus, by means of questions about the circumstances related to the presence and impact of ideologies, we are led toward a substantive understanding of the generation of social patterns and of the part that ideological principles have in that generation, rather than toward an analysis involving “a kaleidoscopic range of clusters’’ uneasily accompanied by unrelated and unexplained idea systems. To conclude, I would like to return to the main findings of this essay and to suggest some broader theoretical implications that grow out of them. The first point is that complementary opposition is found not only on the level of ideas, in the folk model, but also at the behavioral level, in the social process, of a number of societies, such as the northern Somali, the Yomut Turkmen, and the Shah Nawazi Baluch. Second, complementary opposition is based primarily upon a genealogical framework, but in societies putting complementary opposition into practice, the lineage framework is supplemented by nongenealogical principles of residence or alliance such that demographic imbalances, adaptational necessities, and other levels of social process do not impede effective segment solidarity, balanced opposition, and complementary opposition. That these supplementary principles are included in the folk models of these societies justifies calling the folk models “lineage-plus” ideologies. Third, among those societies that claim to base their political organization upon lineage-based complementary opposition, there is considerable variation in the extent to which actual behavior patterns, day-to-day political alliance and support activities reflect the folk model. It was found, through an examination of a number of cases, that the practice of segment solidarity and complementary opposition was associated with territorial instability, with the continual movement, spatial interpenetration, and mixing of the population. The lineage-plus framework provided a set of organizational guidelines and, on the political level, group-based security, to a mobile population which was not able t o use a territorial framework for such purposes. Where territorial stability was characteristic, as among the Bedouin of Cyrenaica, interests developing out of spatial location and propinquity conflicted with interests defined by the lineage-based folk model and thus undermined to a considerable extent the practice of segment solidarity and complementary opposition. Only among the Turkmen, who seem to have manipulated their genealogical model to conform t o the territorial distribution of lineage segments, could interests based upon territoriality guide political process without violating the lineage-based ideology. These points raise some questions in relation to Sahlins’ argument (1961) that the segmentary lineage is an organization of predatory expansion. His position is that the segmentary lineage system is an organizational means of temporary consolidation of a fragmented tribal polity for concerted external action, and that it develops in a tribe that is expanding into a habitat already occupied by another society at the tribal level of
68
AMERICAN ANTHROPOLOGIST
[80,1978
organization. The emphasis here is upon the highest level of tribal organization, upon the reasons and mechanisms for consolidation at this level. One question that might be raised about this view is whether account has been taken of the need in societies with highly mobile populations for organizational frameworks independent of territorial criteria. Given the condition, specified by Sahlins, of long-term use of restricted resources, and the additional condition of high and erratic mobility, there would be ample internal need for a broad organizational framework defining relations and providing support and security. This would account for the importance of segment solidarity and complementary opposition at the middle ranges of organization and the relative frequency of political process at these middle levels in comparison with the higher levels. A second question would be about the effectiveness of political and military as opposed to conceptual consolidation at the highest levels of organization. Given the paucity of evidence for systematic, unified, large-scale military activity in such societies, one might suggest that at the highest levels of organization the segmentary lineage system provides a more passive kind of support in the context of intertribal conflict and expansion. On the military level, it would define interests and loyalties such that internal conflicts could be suspended so as not t o undermine externally directed military activity by certain segments. On a more general organizational level, it would provide a framework independent of territorial criteria for use in defining group relationships within an expanding, and therefore spatially unstable, population. Returning to the conclusions, point four is that in some societies, such as the Bedouin of Cyrenaica, the folk model strongly asserts the existence of a segmentary lineage system, of segment solidarity and complementary opposition, while on the behavior level, in day-to-day political support and alliance, the dictates of the folk model are not followed, the patterns of segment solidarity and complementary opposition not found, and the interests and implications of other principles and considerations, such as territoriality, seem dominant. I have argued that even in such cases, the lineage ideology can be seen to have some constraining effect. Nonetheless, the discrepancy between obvious interests and actual behavior, on the one hand, and the folk model, on the other hand, cannot be denied. The question that must be addressed, then, is why such a folk model is held at all. The answer is that the folk model is not a mirror of actual practice, or even a model for guiding behavior under present conditions. Rather, it is a kind of conceptual insurance, a social structure in reserve, available for activation in the future when current conditions disappear and an alternative organizational form, one not based upon territoriality, is required. This particular organizational form, the segmentary lineage, adapted to situations of population mobility, is especially suited to areas in which political conditions and productive activities result in an alternation through history of stable periods and periods of upheaval, periods of stable territoriality and periods of high spatial mobility and population mixing. To be available for activation as required, such an abstract model must be maintained by the population even during times of stability, though it might not conform to current practice which must be denied or defined as exceptional. The holders of such a folk model have a strong sense of its importance, a feeling that the asserted organizational form provides security for them and that a denial of this organizational form would be a serious loss to them. This is not an example of the cake of custom, of conservative thinking, or of the joy of abstract models. It is an understanding, perhaps more explicit for some and more implicit for others, that this organizational model is an important tool in their social repertoire, and that their environment has previously been such, and probably continues to be such, that this conceptual tool will be needed for their well-being and perhaps for their survival. This folk model, in such societies, is thus not a model of current day-to-day practice, for these people know very well what they are up to. And it is only weakly a model for behavior, since people must operate under current conditions; they limit themselves only to avoiding blatant, open contradictions of purported loyalties. This folk model is a model of what has
Salzman]
COMPLEMENTAR Y 0PPOSI T I 0N
69
been required in t h e past and is a model for what might be needed in the future. It is a social structure in reserve. This final point suggests that idea systems, folk models, and ideologies d o not have simple one-to-one relationships with behavior, either in the direction of guiding and explicating it or in the direction of camouflaging and obscuring it. Ideologies and folk models can have lives of their own, independent of direct and immediate ties to behavior. In t h e case of the segmentary lineage model, it is used to maintain t h e availability of a n alternative mode of organization. As such, this folk model is an adaptive mechanism, available for application should certain not unlikely circumstances arise.4 The moral is clear: even in our studies of the “simpler” peoples of the world, we must not assume that the relationship between ideas and behavior is straightforward and simple, for complex interrelationships and subtle, indirect interconnections between ideas and behavior are just as characteristic of tribal society as they are of the larger, literate societies. NOTES Acknowledgments. An earlier version of this paper was presented in March 1976 at the School of Oriental and African Studies, University of London. Thanks are due t o members of the School and to anthropologists from elsewhere in the University for their gracious reception and their constructive and suggestive comments, and especially to R. Tapper for his encouragement and hospitality. Since that presentation, helpful suggestions have been kindly made by J. Jiggins, J. Rousseau, and J. Salzman. Support for the writing of this paper was generously provided by the Canada Council and McPill University. In a recent paper, Peters (1975) restates his position but does not revise or elaborate the arguments about complementary opposition and lineage organization set out in his earlier (1967) paper. The primary concern of this recent paper is the nature of feud, particularly whether it is a mechanism of social control, functioning to maintain the peace and to redress violations of the social order, or whether it is a self-perpetuating form of disruption and violent competition, based upon underlying conflicts of interest and struggles for scarce resources. This most interesting question is outside the scope of the present inquiry, and no further reference will be made to the recent paper (1975) of Peters. *There is ample evidence that even in the course of everyday affairs, a society will have several alternative structural principles that can be put into effect sequentially, as circumstances warrant. Gearing (1958) has nicely pinpointed this process, calling the sequence of applied organizational principles “structural poses.” He argues that the social structure of a society is the sum of the several structural poses it assumes throughout the year. My suggestion in this paper that ideologies can be social structures in reserve is merely an extension into historical time of the understanding that different circumstances require different forms of organization. This argument, that institutionalized symbol systems can reflect the necessity to adjust organizational form as conditions change and can function to facilitate adaptation and ease social change, is a step toward clarification of the mechanisms and processes underlying transitions in organizational patterns. 3Research was carried out in 1967-68 with support from the National Science Foundation, in 1972-73 and 1976 with the support of the Canada Council and McGill University. I am grateful t o those agencies, the Government of Iran for permitting my research, the Shah Nawazi Baluch, and especially my hosts, the Dadolzai and Yar Mahmudzai, for their unending hospitality and cooperation. 4 B ~ there t are other ways in which idea systems can stand independent of behavior, as in cases where the ideology asserts core beliefs or basic values, even though it does not seem to guide or reflect behavior directly. Examples of the kinds of core values embedded in such ideologies are “our first duty is to God and his laws,” “from each according to his ability to each according t o his need,” “all men are equal before t h e law,” “a country of t w o equal cultures,” and “all men are equal before God.” Such ideologies can set limits o n the degree of deviation of behavior from ideals by providing rallying points and symbolic charters for reform movements and rebellions and can play a part in generating social change by providing invidious comparisons between actual behavior and ideals and by providing an image of more desirable behavior patterns. In the world of the Bedouin of Cyrenaica, Islam has often played precisely this role (Evans-Pritchard 1949).
70
AMERICAN ANTHROPOLOGIST
[80, 1978
REFERENCES CITED Barth, Frederik 1959 Political Leadershiu among- Swat Pathans. London: Athlone Press. Colson, Elizabeth 1962 The Plateau Tonga of Northern Rhodesia. Manchester : Manchester University Press. Evans-Pritchard, E. E. 1940 The Nuer. Oxford: Clarendon Press. 1949 The Sanusi of Cvrenaica. Oxford: Clarendon Press. Fortes, Meyer 1945 The Dynamics of Clanship among the Tallensi. London: Oxford University Press. Gearing, Frederick 1958 The Structural Poses of 18th-Century Cherokee Villages. American Anthropologist 6 0 :1148-1156. Gellner , Ernest 1973 Introduction: Approaches to Nomadism. I n The Desert and the Sown. Cynthia Nelson, ed. Pp. 1-9. University of California Institute of International Studies, Research Series, No. 21. Gluckman, Max 1959 Custom and Conflict in Africa. Glencoe: Free Press. Irons, William 1975 The Yomut Turkmen of Iran. University of Michigan Museum of Anthropology, Anthropological - - Papers. . No. 58. Lewis, I. M. 1961 A Pastoral Democracv. London : Oxford Universitv Press. Peters, E. L. 1967 Some Structural Aspects of the Feud among the Camel-Herding Bedouin of Cyrenaica. Africa 37 :261-282. 1975 Foreword. I n Cohesive Force, by Jacob Black-Michaud. Pp. ix-xxvii. Oxford: Basil Blackwell. Sahlins, Marshall 1961 The Segmentary Lineage: An Organization of Predatory Expansion. American Anthropologist 6 3 :3 22-343. Submitted 1 2 July 1976 Accepted 5 March 1917 Final revisions received 3 October 1977
