Professor Keesing's reply to my article on the authority of ancestorsraises a number of interesting issues. I concur with his stress onconceptualisations which contribute to comparative analysis, but limitmyself here to specifying further a few of our disagreements.

Keesing's analysis remains quite unsociological. By this Imean that he neglects the importance of demography, social relationshipsand the organisation of corporate groups. When Fortes studies theTallensi, their population was over 45,000, growing fairly rapidly,and densely settled on land which grew in value as it was manuredover the generations. As Keesing described the Kwaio, by contrast,'settlements are tiny ... scattered, and frequently moved'(1970: 755). There was only some 3000 Kwaio remaining in the interiormountains for him to observe after the depopulation and migrationsbrought about by colonial rule. Despite Keesing's assertionthat Kwaio social organisation

Keesing's analysis focuses persistently on the categories ofrelationships derived from an egocentric viewpoint rather than fromstructures of social relationships. Forte's primary concernis with group membership. not individual statuses, though he doestend to slip without clear distinction from one point of view to theother. In my article I attempted to stress the sociological sideof Forte's argument: lineages are the basis for structured socialgroups, not just conceptual categories. Jural and political groupsrelate to one another through the recognition of ancestral authority. Despite the introduction of non-agnates (as by adoption), descentprovides the medium of discourse for a central part of social activity. In this connexion, I disagree categorically with Keesing's Schefflrianview that descent groups are a 'realisation' of a conceptualsystem (Man (N.S.), 18, 185-90). Thisassigns much too much primacy to culture, and even at the culturallevel produces a static analysis which obscures the contributionswhich various specific contexts of social action make to linguisticdiscourse. Both language and substantive ideas about kinship anddescent influence the ways in which descent groups are formed, butthis group formation can hardly be explained apart rom social actionand material factors. Moreover, while the Kwaio may make relativelylittle of the agnatic definition of their social groups, the Tallensitake patriliny quite seriously. They do not produce allegedly agnaticlineages in which the majority of members

may be known non-agnates. And Tale sacrifices must pass through specific agnaticdescendants; they are not carried out by an undifferentiated classof officiants (preferably but not necessarily agnates) on behalf ofan agnatic group.

Keesing rather adamantly refuses to look at social groups as such. Thus he takes Fortes and me to task for contrasting 'agnatic'or 'patrilineal' with 'matrilateral'. He notesthat 'a Tale man's nonagnatic ancestors include many thatare not matrilineal (in relation to him): his father'smother's lineage ancestors; his father's father's mother'slineage ancestors, etc. They may have been "matrilateral"ancestors to somebody ... but they are patrilateralto him' (p. 186). Fortes and I may have contributed to somesemantic confusion, but the substantive issue is that however theserelationships may appear 'from ego's point of view'they are important primarily from the point of view of ego'smembership in a corporate group - a lineage - and allthe listed relationships are matrilateral for the lineage. Discourse focused on social groups has a different content from ego-centricdiscourse; it refers to different social relationships. It happensthat Tallensi use the same kinship terms (informed by context) todescribe these different relationships but that should not preventus from grasping the difference. Though Fortes does recognise bothlevels, he has obscured the distinction somewhat by following Talefolk practice (which coincides with his own views) and describingthe relationships constituted at the level of social groups as extensionsof the key interpersonal relationships of the family. He does notindicate any Tale terminological distinction comparable to the Nuerdivision of but (agnatic corporate group relations) frommar (cognatic interpersonal kin relations) but his analysisis still similar to Evans-Pritchard's (1940: 193-4). Thus hisusage of 'matrilateral' to refer to a collective class ofrelationships (matrilateral to a lineage) follows Tale constructionbut may confuse literal-minded anthropologists. Unfortunately, evenour own professional kinship terminology does not have an effectiveway of discriminating the two levels of discourse without the introductionof numerous qualifiers.

This issue lies behind numerous arguments Keesing picks with Fortes(and/or my interpretation of Fortes). He suggests that the nonagnateswho attend a Tale sacrifice comprise an 'array' definedby their common descent status, neglecting to point out that thisarray, however conceptually defined, is not a corporate group butstill a collection of individuals. He tries to argue that Fortesshould have referred to the descendant of a man linked matrilaterallyto a Tale clan as than clan's classifactory 'grandson'(!) rather than 'sister's son' without consideringthat the individual in question was not a member

of the corporate lineage group (Man (N.S.), 18,186; cf. Fortes 1949: 150). In short, Keesing reasonably points outa distinction between kinship and descent, but then unreasonably assumesthat ego-centric descent analysis supersedes socially focused analysisof corporate kinship groups. The definition of descent under whichFortes and I have worked may be problematic; I will accept Scheffler's(1966) point that descent should not be understood as referring necessarilyto a unilineal rule of corporate group membership. Alterations invocabulary will not help analyses, however, unless we still have away to stress which relations define corporate groups and which donot.

As long as Keesing treats kinship as referring solely to ego-centricanalysis, he will have to come up with another term to denote therelationships which organise people in corporate groups, of he willmost likely continue to neglect them. It is no doubt true that conceptualconstructions need to be distinguished from social groups, but thatneed not involve granting them an ontological status independent of(and apparently prior to) social action and organisation. It is noaccident that Keesing's ego-centred mode of analysis ends upwith a special stress on cognatic descent; both his field situationand his theoretical orientation suggest it. It is impressive to recall,however, that Forte's Tallensi monographs took as one of theirmajor points and organisational features the distinction between thesocial structural analysis of agnatic groupings and the analysis ofinterpersonal relationships appropriate to what Keesing would haveus call cognatic descent.

Craig Calhoun
University of North Carolina,
Chapel Hill
I am grateful to Lee Schlesinger and Edward LiPuma for helpful discussionsof this comment.

Evans-Pritchard, E.E. 1940. The Nuer. Oxford: ClarendonPress.
Keesing, R. 1970. Shrines, ancestors, and cognatic descent: The Kwaioand Tallensi . Am Anthrop. 72, 755-75.
- 1983. Ancestors, sociology and comparative analysis. Man(N.S.), 18, 185-90.
Scheffler, H.W. 1966. Ancestor worship in anthropology: reflectionson descent and descent groups. Curr. anthrop. 7,541-51.

Further debate here on the Tallensi would be unproductive. However,on the modern Kwaio, Calhoun is simply wrong. I was able to reconstructa highly detailed and very accurate picture of Kwaio social organisationin late 1927, on the eve of pacification (see Keesing and Corris 1980;Keesing 1983).

These data, and twenty years of ethnographic evidence, allow detailedassessment of continuity and change in Kwaio social organisation. While Christianisation and the 1927 punitive expedition did thinpopulation and leave empty spaces (Keesing & Corris 1980; Keesing1983), there is clear evidence that in the 1920's (if settlementswere tiny, scattered and shifting (1927 mean population 9.95; 1963mean 8.9): (2) descent groups were tiny (3-15 adult men, in 1927);(3) these descent groups were contextually defined and corporate onlyin limited contexts (Keesing 1971); and (4) cognatic descent and kinshipwere crucial at the level of individual and intergroup relations. In 1983, both social structure and ancestral religion remain substantiallyintact: reports of the death of Kwaio society are greatly exaggerated.
Roger M. Keesing

Australian National University
Keesing, R.M.
- 1971. Descent, residence and cultural codes. Anthropologyin Oceania (eds) L.R. Hiatt & C. Jayawardena, Sydney: Anges& Robertson.
- 1983. Kwaio religion: the living and the dead in a SolomonIsland society. New York: Columbia Univ. Press.
- & P. Corris. 1980. Lightning meets the west wind: theMalaita massacre. Melbourne: Oxford Univ. Press.

This correspondence is now closed - EDITOR