Mb - /5 i./ Rodney Needham Remarks on the Analysis of Kinship and Marriage /f~fc What is theory in musical composition? - Hindsight. It doesn't exist. There are compositions from which it is deduced.' IOOB STRAVINSKY iil hii INTRODUCTION To consider the analysis of kinship and marriage leads at once a paradox. ^'Kinship is to anthropology what logic is to philosophy or mude is to art,' declares Robin Fox; 'it is the basic discipline fpf the subject' (1967: 10). Certainly it is a well-established part 'the subject: Lafitau began the comparative study of uni-neal descent and classificatory terminologies as long ago as |1724, and Morgan inaugurated in 1871 what has since become a ecognizcd topic of academic investigation and theory. The fiyllabuses of university instruction in anthropology now invari-mclude kinship; no textbook is thought adequate without pome tieatment of it; and in all the variety of examinations in pnthropology it occupies a central and unavoidable position. lYou cannot gain a certificate of competence as a social anthropologist without demonstrating a command of kinship theory, it is expected of most anthropologists that they will make jfBome contribution to it. Indeed, most of those who have made fgreat names in anthropology - e.g. Rivers, Kroeber, Radcliffe-3rown, Lcvi-Strauss - have gained their prominence largely ' their publications on kinship. If there is one topic, therefore, Iwhich is indispensable to social anthropology, and which defines Iwliat social anthropologists essentially do, it would appear to |be kinship. Here, if anywhere in the subject, we should expect Jto find discipline, methodical rigour, and theoretical advance. This much is, I suppose, a standard account of the matter; »but an inside look at what really goes on reveals a curious 1 M m i«? m Jiodncy íVcedham gsM situation. The majorjty_of s^dent£_gXiHlltoop^[lo^ijmd_^^S they have as little to do with it as they can get away mOkI Examination scripts seldom show much enthusiasm or soúnrai knowledge, and the professionals often seem not to be partio||i larly good at the practical analysis of kinship systems. Théři^S a comparative paucity of published works on the topic, án|| progress in understanding kinship systems has been sporác^ and slight. The current theoretical position is obscure and coifs fused, and there is little clear indication of what future develop ments we can expect or should encourage. ;$||Í In view of the constant professional attention extending oval roughly a century, and a general improvement in ethnograpmjl accounts, this is a remarkably unsatisfactory situation in wh|| is supposed to be a basic discipline. Obviously, after so longís time, and so much field research, it is not just facts that.™ need. Something more fundamental seems to have gone wrong! What we have to look for, perhaps^ is some radical.jBawJrai anaTyšisT^o^uTTrutiaTHělect in the way we approach the pheriiaS __^ m75ŕTirEa1ňífu^^ his salu^^°ä^dTe^s*^eth^nldng Anthropol^y^^rflST-'TŽ?^ IJüTinörelFTIFEIe^igfrEhät even hislŤeiveTčľarity, and ingenuityfl have yet had much effect on received ideas and ordinary práo|| tice. What I want to do here, then, is to resume Leach's icon$|i clasm and to look with him for a way out of our presenfi uncertainties. ||| A possible diagnosis may be that the trouble lies not so muchl in the substantive study of institutions of kinship and marriage! as in our conceptual premisses, and most decisively in the way! we conceive the classification of phenomena. The failings! basically responsible for the present situation are, I shall argued firstly what Wittgenstein has called 'a craving for generality^ (1958: 17) and secondly the lingering delusion of a naturalj science of society, a conception which has led to a kind of ana|| lysis that has produced few useful results. You will already^ recognize in these contentions an echo of Leach's strictures onl 'butterfly-collecting' and on biased premisses. For that matter;!! what I too want to press for is precisely to 'take each case as it|? comes' (Leach 1001: 10). But I think there is a conceptual>$ therapy by means of which we can prepare ourselves better to|5 Hemarlcs on the Analysis of kinship and Jlunuuje. ijto,iand for this purpose I suggest that we should turn back Jiiowie's Culture and Ethnology (1917) and to Wittgenstein's ÍM^Book (written in 1933-34, published 1958). You will see éŕéfore that I cannot pretend to be telling you anything very WBut since it has taken me á long time to see the relevance Sithe useful effect of views published decades ago, I suppose efěfwill be others to whom it will be helpful if I rehearse 'mion this occasion. "y argument is presented in the form of remarks on a series topics, not as a progressive exposition, and without any large apparatus of scholarly and ethnographical references. I adopt (means for two chief reasons. The first is that I have already "Wished enough work on kinship and marriage to excuse me Tesěhting any detailed demonstration of what I think ought ibe done by way of analysis. This permits me to make my Spihts in a cursory style which may stick more readily in the *éiňory. The second reason is that a fuller citation of pro-Buncements on kinship uttered by many of my colleagues xjuld have to be rather dissentient, whereas my intention is ^|be positive. For the most part, then, I shall cite only those *th whose opinions I agree, anrTnöt e^ň^mäný'of these. The lisentffirís~ňBTto~tax you with facts or with academic controversy, but to concentrate on the concepts that we are professionally inclined to employ when we analyse institutions of Jnhship and marriage. I shall deal, rapidly, with the notions of 'hship, marriage, descent, terminology, and incest. ii KINSHIP There has been a fair amount of discussion about what 'kinship' ■really is. My own view is that much of this debate is pretty scholastic and inconsequential, and I shall not recapitulate any »of it or embark on yet another definitional exercise. Let me .simply adopt the minimal premiss that kinshipJ^a£_to dojvith %he allocation of rightSluŤ|ro gHolíníoTJiěTielčtTTliěšěrights are not of any specific kind but are | íěxčeěThňgTy^vořious: they include most prominently rights of Jgroup membership, succession to office, inheritance of property,! 3 liudiHHj A ccd/tiim í/ locality of residence, type of occupation, and a great deal else, They are all, however, transmissible by modes which hav| nothing toUo withľthe sex^örgenealogical status of transmitter OljiJčhä^ connexion with the; facts, or the cultural idioms, of procreation. It is true that the! possession and exercise of these rights is defined by reference tô the sex of the persons thus related; but then so is the division of labour in the simpler societies, yet we do not for that reason think this method of distinguishing statuses so remarkable aá to deserve a special designation and to call for a distinct typa of theory. '£ These jural systems and their component stotusesjcajťb| geneáTogičliilIyl^^ a fundamental' __%, question that has never been prop^řTy"řěšST?ě37luToTTcaňňbt talčeTlHřpTíereTTe^^ is certainly a' very convenient fact, but the method of description does not entail any particular property in what is described. Thejar; cumstance that two societies can be described by the same means dügnt^ either jwiologioally or semantÚ2aJ^yJ_b£twee2ithem. Still less does it mean that thejg relationship so"| Cuaceiyed by the actors. ,' What information is given, then, by the report that an institution has to do with 'kinship'? Nothing, really, aboutl social facts. For the label designates no distinct type of pheno-Jl mena; it provides no clue to comprehension; and it does notji indicate the kind of analysis that will be appropriate. The usej| of the word 'kinship' is to be found, rather, in the multiple con-|p notations of common usage, in the organization of ethno-a graphical accounts, and in the conventions of academic dis-,|| course. When an ethnographer gives one chapter the heading^, 'Kinship', and another the heading 'Sacrifice', we have a rough || preliminary idea of the different matters they will describe.'j§ It may well turn out, though, that there is a close connexion' between them, just the same, and very likely neither will be comprehensible without the other. Similarly, if a colleague tells you that he is interested in kinship, his choice of phrase implies, that he could have stated instead that he was keen on sub-, sistenco economies or primitive law, and the word he actually employs does indeed give you a vague idea of his theoretical iA*i;tv ■,_ llemarks on the Anulysis of Kinship und Jlarriaye ' ôňfc, the books he has presumably read, and the kind of tech-íoaL conversation he is likely to engage you in. In this case as roll/ however, it cannot be inferred that his interest in kinship jill'be unconnected with economics or law; and in fact, of jjeourse, it will probably turn but that he has to deal with these Íí<§nós also and that they in turn demand a recourse to kinship. |Iiam not denying, therefore, that the word^KnSfiTp' is useful; oleí[ still less should I wish to try to reform our professional rooabulary by narrowing the definition of the word or, on the Mother hand, by urging that it be abandoned altogether. What Síam saying is that it does notjienote a discriminable class of phenomena or a distinct type of theory. We are tempted to 'think that it must have this specificity, because it is a substantive and because it is an instrument of communication. [But it has an immense variety of uses, in that all sorts of Institutions and practices and ideas can be referred to by it. Segmentary organization, section systems, widow inheritance, ^polyandry, teknonymy, divorce rates, and so on - all these Jfcppics and very many more can be subsumed under the general Krúbrio of kinship. In other words, the term 'kinship' is what ^Wittgenstein calls an 'odd-job' word (1958: 43-4), and we only čget into trouble when we assume that it must have some ^specific function. In a way, it could be said that the trouble is J lot very serious, since when we actually investigate an insti-> iition, or compare ways of explaining it, we do not generally ppeak of kinship at all. Indeed, this common circumstance ^demonstrates that the word has in fact no analytical value. jp)n the other hand, anthropologists do often get into trouble, pÓfJa timewasting and discouraging sort, when they argue »about what kinship really is or when they try to propound pome general theory based on the presumption that kinship lihas a distinct and concrete identity. fM> To put it very bluntly, then, there is no such thing as kinship, Sand it follows that there can be no such thing as kinship theory. in MARRIAGE ítCJfu^X- i^H.ýLtn bi^u^u-f- fv ?>Very similar considerations apply to the concept of marriage b'and to the theoretical propositions of anthropologists about liiHnuy i\ LCti/íiiiii ,, j marriage. I need not say much jibont this topic because'íifif case has been well made byíeacp 'íP^riÍM^s^_i-í-l!äJ?]?]Í(l'Í of rights": hence all universardefinitions of marriage are vaM (1001: 105). ;>$|| I think there is no refutation of this argument. What I shoúll like to re-emphasize, simply, is Leach's conclusion that.'tliP nature of the marriage jjtts^irjfciojiJiípS^^ principles pXjdeigent arid nijes of resjdence, (108). Perhaps':}! is not so much correlated, though, as it is defined in any pari ticular instance by what we divisively call the 'other institujl tions' of the society. It is not only jural institutions, eithers that we have to take into account, but moral and mysticálj ideas as well, and these in an unpredictable and uncontrollable! variety. The comparison of marriage in different societies needs! therefore to be contextual, and ultimately 'total' in a Maussianjj sense, if we are to bo sure that we understand what we are trying! to compare. , im In this connexion, the designation of marriage has a specials interest. Ethnographers do not on the whole report the indi-1 gcnous terms for marriage, or investigate the connotations ofe such terms, yet Ave need not look far to see that these can beJ revealing. For instance, the modern German Ehe derives fronij MHG l, ěwe, law, statute, and its recent narrower meaningf merely singles out marriage as one of the most importantjurall inlillrüfiöTS^^ how-1! ever, come from the Latin mimlus, nusbanaTwmch is usually^ referred to IE *mer- *mor~, represented by various wordajj meaning 'young man, young woman'. It is at once evideptrthatjij we^TtwTTTiliiro^ twojp;utojlM^ this, there mayTiöTl3e~7i^^fesign~ati^ at all. Inil classical Greek, as Aristotle observed, 'the union of man andl! woman lias no name {roMics,J^3L2jJJbJvGn though marriage J was essential for tTíe~přeservation of the 'houses' (oIkoi), m which were the constituent elements of the Athenian city-state,if there was no single word which could be taken to stand for1^ 'marriage' - nor, for that matter, were there words in classical % Greek which stood for 'husband' and 'wife' (Harrison 1908: l). í And to take a contrasted enough civilization, whereas the ■$ Penan of Borneo do have words for husband (banen) and wife'^ 6 I Wtti'rfo), they too have no word for marriage. One wonders, irefore, how many other societies make no lexical recognition Jhat institution which has so commonly been regarded in jflrŕópology as categorically essential and universal. isoon, however, as we adopt some technical definition of Image, whether or not it is held to be universal, we run the ijóf' leaving out of account precisely that feature (e.g. itity, allegiance, life-giving) which in one or other of the iiěties compared is in fact central to the institution. This is i|cburse a familiar quandary in comparative studies, but I "ink"it is a question whether its lessons have everywhere sunk (T-;home. At any rate, large-scale correlations are still mpted, and these can be carried out only by means of ly strict definitions which are nevertheless presumed to be dely applicable, but the stricter they are the less likely it is at they will cope adequately with social reality. Once again, though, I am not denying that 'marriage' is a useful word. On the contrary, it has all the resources of meaning which its long history has conferred upon it, and we Bnould now find it hard to communicate without these. For ,hat matter, it is a more indispensable word than 'kinship' is, it directs us more precisely to an identifiable kind of relationship. If an ethnographer sets out to tell us about mar-, we have at least a preliminary indication that he is not ■going to focus directly on dam-building. But I choose this latter eiample, all the same, precisely because Onvlee has shown that fir!eastern Sumba, wb^re^marriage lš~^eFčrTrľe3T"witli^the mátrilateranirčräsj^^ tion of ďam-building unlessjraujfirst understand the norms of "A^ffigejniS^IISlí- Conversely, you cannot understand the marriage institution without knowing the forms of co-operation which follow from it. There are also cosmological grounds to both aspects of Sumbanese social life. As Hocart says in ^another context, 'There is much more to the cross-cousin system than the classification of relatives; there is a whole theology . . .' (1952: 237). But nothing in the ordinary connotations of the English word 'marriage' prepares us to grasp a situation such as this, and nothing in anthropological usage gives the word any technical value either. b So 'marriage' too, is an odd-job word: very handy in all 7 w fti&h .liuiuiť.t/ iVix.dliuiil sorts of descriptive sentences, but worse than misleading^ comparison and of no real use at all in analysis. ■'&& •wt C i-U'íu-'c'L- ^í- ~ 5ří'»!i '••*v.'i-, m m m. m Ú H Srb*, r} í> ,-f 1: i-Ä. /. y *£ DESOE #r ZW-.- < |ä -< ty éS~s£~* 1 The classification of modes of descent is a specially effective examplo of conceptual difficulty, because the topic has been! constant anthropological concern since Bachofen and McLennan In spite of this prolonged concern, however, there is still nôf general agreement on the matter. Anthropologists habitually use terms such as 'patrilinealVor 'matrilineaľ, yet cannot easily claim that these are specific d^čTiptionsľ Even "wďlielľTJuTlRlňl^^ dispute, it is sometimes possible to argue about the type off descent system to which a given society should be assigned^! Or when it is agreed that a society is patrilineal, forexample,f it is possible to argue about whether it is a strong or a weak! ^^^cěofJJiaiype. Such arguments might be all to the good iff they led to cogent and agreed decisions, but for the most partj this is just what they cannot do. What we are left with is not! theoretical advance but a wordy conflict of rival definitions.; I need not go on, or supply examples, for the situation is familiarl to any anthropologist. What is important is to find a way out off this typological confusion. S°me ü£ui!!E2E2!08'sks (e-6- Kokken» Lewis) have contended! that wo cannot assimilate different societies on the mere ground! %jja|TljoyTžŤ^^ tojjcjjrolčelíll^^ ófřights thät^ whier _____________________________......____.....__ f cwe| tlie respects in which societies are to be counted as patrilineal andŽ thus as comparabTo. Ihe iurujlioiuTnTnTpTicfitions oTdeščeht arej often much more significant than whether descent is traced inl the patri- or matri-line' (Lewis 1065: 109). This approach has| a well-established ancestry - it was embarked on by Fison in J 1879 and continued by Wake in 1889 - but it is not, I think,!} the answer. The difficulty remains that this substantive concentration on complexes of rights makes comparison as uncon-'| trollablo and as hard to carry out as the rights may be various 8 Remarks on the Analysis of Kinski}) und Alarrituje, y to specify more narrowly what 'patrilineal' means in a ioular description does not make the rubric any more fill; in fact, it tends to show how inappropriate it really is. jthe other hand, no one neetls it in order to define a single ijral;system that is under study, for an exact survey can report efsocial facts without summing them up under any such eral label. In any case, the proposed reconsideration is not damental enough. Lewis, for example, still takes it that ascent' is traced in some line, only that this tracing is affected vt'other principles of organization' (1965: 106). Yet it is this "ry;notion of 'descent' that needs investigation. Ajmore radical course is pointed by Leach's suggestion that óh typifying devices as 'patrilineal' and 'matrilineaľ may haye '^sociological significance whatever: it may be that to create a class labelled matrilineal societies Jf?as irrelevant for our understanding of social structure as -the creation of a class blue butterflies is irrelevant for the (understanding of the anatomical structure of lepidoptcra' 1961: 4). 's is a bracing notion, and in fact it is far from clear that there *íahy convincing defence of the class of matrilineal societies. y.own inclination is to doubt whether there are any useful repositions about matrilineal systems which distinguish these, Ma class, from societies Avith other rules of descent and thereby ustify the typology. This may be thought a matter of debate, ut.at least there is enough reasonable doubt about the issue xall the conventional typology into reneAved question. :Yet can Ave do without it? After all, Bachofen and McLennan ěated a stir because they had discovered something; and the jural differences between the Minangkabau and the Batak are al, not the result of an unfortunate typology. The troubles egin, however, Avhen we try to characterize and convpjure_thc ®ňlmjp!äbävl^^ they 'arejjojijijnatrilin^aj^in other words, when we. extend the char? aoterization beyond those features whichjgrompt the descrip- Iňň~1'maTärflró^ labeTap^Tíěa^Io airthT^th^Flnstitutions also. Here I think we have a double ^'iiíMwočě~~^r^íě~~cmvhig for generality: Ave classify societies together because by some definition they possess in common the 9 feiiturc of 'matrilineal descent', even though the riglitsä governed andTnTě~TTměftoTTs~ftTe^^ and in each enso we oJaliinřyTnl!t^^ á^matrilineal society', because väHöl^cTnirTiiäges, which'mB not 1)0 commolnfiropéřties of members of the class, happei™ accord with that mode of transmitting the definitive righils Clearly the method is wrong, but all the same there is so™ thing to be wrong about. What is the solution? There is*?® wayout, I. „think, namely to resort to purely_forjmxj_crjter)l i.e^to ceasej^n^n trjijn^ aggregates of many kinds, exercising different functions, ail to think instead in terms ofjogicai possibilities. I am notH! ferring to tneTuse of systematic models, but to a far more báští procedure. It may strike you as excessively simple, but I thinS it has advantages. ;r|jj| Given two sexes, and transmission of rights defined by thcJ| we may distinguish six elementary modes of descent. Letftjf denoto male by m, and female by /. The modes are then™ follow: jJs ' ms Jßit, 12./-»/ m 3. (m -mi») + (/->/) M 4. (m->f) + (f-+m) M 5. (m->m) || (/->/) '1 0. mlf-*mlf Jj These formal modes correspond severally to functional para« digms. Mode 1, male to male: patrilineal. Mode 2, female t$| female: matrilineal. Mode 3, a combination of modes 1 andtíl in the definition of any status: two conjoint rules of descent]! i.e. bilineal. Mode 4, male to female, female to male: alternating;! Mode 5, male to male, female to female, defining distinct! sexual statuses: two disjunct rules of descent, i.e. parallel!! Mode 0, male or female to male or female: cognatic. f||j Naturally, once alternation is admitted the number of adoMj tional possibilities (i.e. variations on mode 4) is increased, bull these six modes are elementary. [m The modes are not to be cojj^jyj^jj^ charactejjzjngjocietie|| hoffstlcally. Indeed, modes 4 and 5 probably could not béf employed socially as regular and exclusive principles of trans-l 10 1 !on; and incorporation, though certain rare and uncertain ŕôximations to them have been reported. My point, rather, Sat in any society different rights may be transmitted accord-ťofdifferent modes. The ethnographer's task is then to sort the'rights, according to th)e indigenous classification in the ipiace, and to establish the various rules by which they are frned. To cite an extremely simple case, Penan society could äŕtly analysed in this way: descent name mode 1 residence mode 2 inheritance mode 5 group membership mode 6 (result is thus a complex of rights and rules, not correspond- "entirely to any single principle of descent. The jural complex j'arrived at is likely to be singular, not only in the kinds of 'flits culturally distinguished, but also in the association of ese with different modes. Only in the extremely improbable event that a society were to träniinlt all rí^iiľuiňToTmlyTilfonir jngle mo£e^mM it^^ lit', would of course be possible to construct a new and much xtehded typology of descent systems, composed of all the logically possible combinations of the elementary modes. This Jroúld give a total of 63 types. Whether this was worth doing would depend on the results, but since my present concern is tö-undermine such typologies it is not ElTmidertTťktngTfiäTT BKôinôrrecímTmeTrdľTh^firtention is to preserve the specificity öftthe social facts, yet at the same time to make possible a comparison that is not based on merely contingent assemblages institutions or 'functions'. This we can do by resorting to those simple logical possibilities which govern equally both the practical fabrication of social systems and our own abstract conceptions of the forms they can take. That is, our analysis will be guided by the same logical constraints as must have been effective in producing the systems that we study. %:Of course, there is still ample room for the arbitrary or mistaken discrimination of kinds of rights or modes of transmission, »fas well as for speculative alternatives in the selection or classification of those rights or modes that an investigator is #E 11 M interested in. Wc can never obviate these sources of ímprociaiói and variations, but at Jeasfc a purely formal approach give^j the chance of making a less biased start. '.:.M Let me admit afc once, moreover, that this has been a cursqi exposition, and that I have not distinguished exactly, as migl ultimately be done, all of the jural variables involved. The ra reason is that I am not trying to work out a technique, compl with instructionsnroř~iiiě7~bTiFto bri"" nU—i conception of descent systems m^TľTnTmTeT^TálčIväňTägešT" ,/iv 1. To begin by listing logical possibilities, without regard jfc their social feasibility or known realization, actually acco better with social reality. TJyojniligeiiojis^classifie^^ canJ)e_jh^recJly__ajlo4Ui^, and these can then be distribui without prejudice among the formal modes, not in accordaii with some theoretical predisposition or academic fashion.TJ approach thus conduces to an accurate ethnographic descri] tion, a possible benefit which is surely of the very greai importance. 2. Itjremoves the jgmjptajjon to characterize a descent .._ iinivocäTTF^sTpätrilineal or matrilineal, etc., and in doing make a biased choice of any one kind of right which is then to define 'the' rule of descent. ii| 3. It renders less easy, consequently, a comparison of different] of jural and other diíTerfiiTv>° ™u:~t- • ____»«iní m a descent group, ančTtJie ověřnding| of jural and other differences which might qualify or invalidate' tho comparison. 4. A logical analysis presents a constant discouragement to thel employment or elaboration of defective empirical typologies! which have proved not to advance our understanding of thej phenomena. Jj ß. The formal ttpErgaghjgjlo.ws anthropologists of any theorgti-4 cal persuasion to collaborate, or to understand each other, with-! olírOh^~člíštoniäry óbšEäčTes of variant technical vocabularies 5 or opposed connotations implicitly attached to conventional; labels. 12 systen so Hematics on Ute Analysis oj kinship and Jlatnaye Rň|óon8equonco of conceiving descent systems in this way, \ galOne that has a special analytical value, is that among a imbér of societies compared in any formal respect there will not Jasumed to exist any empirical feature common to all. It ^more readily be seen, in other words, that they may not ipófle a class in the conventional sense. Instead, they can ibnTwuiFTmgTiir^^ a simple (mfräílóTíTTi^ by See features (p, ..., v): B r, s, t G t, u, v piť; r and t be each a type of right transmitted in mode Í. there is then a resemblance, r, between A and B, and another, ^between B and 0, but none between A and C. Yet in ordinary ihthropological practice they could all be classed together as patrilineal'. A crucial misdirection can thus be given to our bought by the uncritical employment of the received idea as ŕwhat a class is; whereas to analyse and compare by reference ..[formal, rather than empirical, types can lead to a clearer lew of the distinctive features by which similarities and dis-tttilarities are gauged. piijmm, I am suggesting that the present theoretical con-JDttBion about descent has its origin in two basic concejvtiiaj |||iiects7^ and (2) a pub^cnptlon to a conventional but unrealistic idea of how ja class is formed. These are matters to which 1 shall revert below ^^fJiediscussion of other topics. For the present, I must introduce a note of reserve. A directconsequenceof tho_flEPJoaoh Spät I amjidvocating is*Tn^*cômiparisoriibeecnnes far inorei ^^oui^andon^nyTarge and detailed scale pe^TapTmipractic;! pndějfor CRlTalstnbTino^^ increases Ithe factors in question and does not make the rights thus řaiseriminated any more comparable, , ^ J l Ľ forcefully to the question of what the 'Omaha' label is supposed| to tell us. The answer is that there exists no useful generaliza-?? tion about this factitious class of Omaha terminologies. It is* true of course that the minimal equation is associated withís patriliny (in a conventional acceptation), but no one could Avell ý 14 r-4 Malia Jraaha Jhlithat this was a theoretical result arrived at by means of Ppype. It has been well known, at any rate, since the last iwiry, when both Köhler and Durkheim made the point. Rfit is surely not surprising that the terminological identifica-Ssbfa man and his son is accompanied by a transmission of Pitis through males. If, then, an ethnographer reports an terminology, he tells us nothing of any descriptive fpej:and in representing the terminology to himself under flilabel he tells himself nothing of any analytical value either. |systematic comprehension is thereby provided, nox_dfifiS tKě_8od6tyjwhjolL£mB!oj^^ really is thing as an Omaha termjnotog^ejee^ themselves, and it leads only to confusion and wrong icTuiiönFtösuppose thatJžheje is, -————________ % par alleTcaseis that of the so-called 'Crow* terminologies, Iically defined by the minimal equation FZ = FZD. There many very different forms of terminology which possess this jure, and they do not compose a systematically definable is. A telling illustration is provided by Eggan's paper on orical changes in the Choctaw kinship system (1937). He ins by conceding that 'if we examine various Crow . . . ship systems we find a series of variations . . . , so that there jme difficulty in deciding whether a given kinship system is row type or something else' (35). The cases that he compares íonstrate this point. The Choctaw terminology varies from Crow type in the curious features FZ ^ FZD and FZS ľZSS. The Chickasaw classification differs in other particular», the Creek from these, and the Cherokee yet again. Finally, jjřthe Yuchi terminology, which shows still^jttíi£x_diyergences gfrom the Crow type, is thought to suggest áľCrow systemjeeause £of the inferred equation (FM) = FZ. Buť*The^eŤeřnnnologies, Ifin spite of all their differencesupne from the other, can nevextha> |less be classedjfcojjej^^^ ^anthropologists should wish to do so. tf- Eggan's own purpose is primarily historical: 'These kinship '"structures,' he writes, 'originally Crow in type, were pro-'gressively modified by varying degrees of the same accultura-'tional process' (47). This argument is convincing, but it serves' i at the same time to impair the concept of a Crow terminology. If 15 AW C K m Kvuii though these terminologies can apparently be fcrac back to forms of classification which more nearly resembl each other, in their reported condition they are reallyljyj different - yet they are still typed as 'Crow'. And even jOk h aye undergonea similar k\j^dj)f^iang^ut^^mn^r-piM u res, tli e degrees of variation winch thejj^exhjbit jnusťjjBdl| aňyTl^ol:etičäT^ífecOirclííšsing them jďl as Crow terminolog! ]?o7TíIiě-plmTFTn"l^^ a class of pjienomena is to* able to formulate jiropositions which hold for all members! the class. But in these cases the terminological variations^ argued to correspond to different degrees of social changeM other words, however similar these societies may previous! have been, they are now disparate in certain institutional,!^ spects, and these differences are ignored when they are sut slimed under the 'Crow' label. On_the other hand,jthe labe| neither essential nor helpful in the sociologieal analysis of eaik of the societies which are thus classed together. ;}|i I have cited Eggan's investigation precisely because íthag ethnic connexions and other similarities of circumstance malcM a comparative study especially feasible, but even in this cas^ the conventional typology of terminologies has no comparative or analytical value. This conclusion becomes yet more clear when a comparison is extended to societies in other culture areáli e.g. to the Mota of the Banks Islands or the Ramkokamekra^oS Brazil; for the Crow designation then refers to nothing morel than the common feature of matrilines, which is exactly thfl basis on which the class-designation is applied, and there is riöl further sociological property which can be ascribed to members! of tho class. .. ;;if| Similar conclusions couMjreaddy be reached if wcmea^-tnl jjcnitmjzo_other conventional types of relationship terminology^ e.g. 'Dakota', 'Sudanese', 'Eskimo', 'Hawaiian'. In each instance! itis possible to demonstrate that the class has been imřaMÍpll CfiUSMíllíědLi111^ m each instance, so far as I can judge, no pro|j positions of scientific value have been arrived at by means of the.a typology. >m. What I wish to propose therefore is quite seriously that thisjlj kind of typology, i.e. one in which the types are defined by||| isolated features of named societies arbitrarily selected as5«! paradigm cases, should be entirely abandoned. It is method-,| 16 >/t: ft i. tin jj|faulty,. it misdirects research, and it has served no useful sarinot even say, as though in recognition of a theoretical lüice now accomplished, that the time has come to make this loěptual change, for what Irecommend is no more than what ^pressed for in 1917. Our only advance since then, as jttsithis issue is concerned, has been in factual knowledge, ppiat we now possess superior ethnographic resources for the "loristration of his points. Lowie argued, in his neglected but iterly work Culture and Ethnology, that any given system is pomplex historical growth that cannot be adequately defined ^Přhole by some such 'catchword' as classificatory, Hawaiian, ifhat not (1917: 116). 'It cannot be too strongly urged,' he iphasized, 'that a given nomenclature is molded by disparate jfnciples' (122). 'There is no Hawaiian system, no Dakota flEjm' (123). jfHow, then, should we analyse relationship termmplogjeal,. Jidoning types and conc^ejritrating on principles. 'Wje inilup^ategoriesTíeatures, or principles of classification iah of types of kinship systems' (105).~~~ -———— ^örmypärt, I KavíľtriedToTnake this point in a number of lláces, so perhaps I may simply refer you, for a recent demon-fmration, to my analysis of the Gurage terminology of social [classification (1909). This terminology includes the features ľ = MBS and FZC = ZC, so that by these criteria it clearly 'belongs to the 'Omaha' type. But the really interesting thing J&tbut the Gurage terminology is that it is a quite singular form míclassification, particularly in the lineal equations in tho line Msreference, and systematically unlike any other that I have ítíeen able to discover. I have tried to show in that paper that $he distribution of the terms can well be elucidated by a [sociological analysis of Gurage institutions; but at no point in [the investigation is the Omaha label, or any other aspect of |the typology to which this designation belongs, of the slightest pse. Instead, the analysis proceeds by an attempt to understand Gurage social classification through the categories and ^principles proper to the terminology itself. The most prominent ^principle of this classification happens to be exhibited in the V extensive lineal equations that are effected. But this feature 17 Jí llUllt I does not constitute a type, or assimilate the Gurage terminoloj to the Omaha or to any other type. It is simply one culti exemplification of the fact that there are only two form! possibilities in the terminological identification of statusesin? descent system, viz. lateral and lineal. A terminology may recoj nize either dimension, exploit either possibility, in a variety,! ways. AH that the Gurage have done, by their lineal equation, is to make an unusual exploitation of an elementary prinoiplj of classification. What is interesting, then, is the specific coi figuration in which a terminology takes advantage oftk possible dimension of categorical order. The reasons for whic), it distributes certain categories lineally can only be discovered! I have argued, by an intensive examination of the facts oft! particular case. A recourse to principlesjijid^o^flrjoiaJLPOj!;, bilities enables this to be done; but to place ajjerjninotogjnmdéi a substantive rubric such as_'Ornaha'. in the conventional) *iypölögy7serves no such purpose. We mustjndeedtakejjaeh] cliiFliTlFcömesjjonly I would add that the way we take'il can be guided by formal considerations which help us to recognize* more clearly the intrinsic characteristics of the case. i:\ This example introduces the stock question about the connexion between terminology and social forms. It will support] the general burden of this paper to make just a few remarks this topic. There are presumably two main purposes in trying to establish correlations between categories and action. One is to permit such inferences that the analysis of a particular society shall bo facilitated: given a certain distribution of categories, the! investigator will then know what to expect in his study of;, institutions. The other is to permit sociological comparisons:'! given similar terminologies, it can then be assumed, for the sake of whatever theoretical issue is under investigation, that' the institutions will be similar also. •?£ The sad fact of the matter, though, is that neither of these]j conditions obtains. The outcome has been generally discourage ing in these regards; but let us review the situation and see what lessons nevertheless emerge. i t 1. Relationship terminologies can be divided formally into (a)1 ''ossifications in whichthe typical feature, in the medial three ifieaTöglcal levils^iliistlisľthedistinction of statusej_ac£ojDdL_ jm;ö whether relationship is traced through persons of the tajenej not.~^ ^statuses into descent Jines, such that, e.g. F = FB, S'= MZS, S = BS, as contrasted with F ^ MB, FBS ^ S;, S # ZS. Examples are lineal descent systems such as -Kaguru, Mapuche, or Nyoro. Non-lineal terminologies have 5|such_pj3sitwe^atu^ is|principle of distinction. Examnlej^^ pH as the Penan, Sarakatsani, or Siwang. 'his is such a simple and fundamental division of forms of 'ossification that here, if anywhere, we might expect to find a correlation with institutions. There are indeed certain immon associations, but it is never safe to make sociological nces from the presence or absence of the distinctive eature. A lineal terminology does not entail a lineal rule of does plerer itih {"descent, Tmčľ~Itnr less doě¥TF~ěntin~the presence of lineal JroujiOSxampTeTöf so1čiě1ífěš"wltTíTiněaTtěřminologies íaesčenTi j$ufc without fixed rules of unilineal descent or corresponding fdescent groups are the Sinhalese of Pul Eliya and the Trio lofSurinam. On the other hand, a non-lineal terminology can b^accompanjed by a lineal transmission of ceřtiiň~rigKíi/ succession to office or inheritance of real property. Sxamples are Bali and rural Japan. These empirical conclusions accord with what I have suggested above (sec. IV) about š&'descent. 18 Whjre^^^njljingal terminolog^ca^eguations^either direct "(e.g. MB = MBS, as among the Gurage) or alternating (e.g. 1MB = MBSS, MBS = MBSSS, as among the Iatmiil), we can |be fairly confideniJlifljLsge_shall find in the sphere of institutions ^Bome explicit expression o-f a mode ofdescent (in these examples, ^patrilineal). But this can never be an absolute inference, as we řhave seen in sec. IV above, for there is no telling to what extent i'rights of other kinds may prove to be transmitted by different J modes, and there is no a priori scale of evaluation by which it :>might be presumed that the mode exhibited in the terminology was the most important. i For that matter, it is conceivable, and may in one alleged 19 iiiiii/ít.i/ u\ 1.1 mm m oaso be so in fact, that a terminology composed of matrilini should govern the affairs of a society that was preponderant patrilineal. In any case, even if a rule of descent couldäbl inferred from a terminology, nothing else could be. That|j| one could not infer what groups were formed, how rights wm ascribed, or Avhat values were recognized. 'f® 3. Prescriptive terminologies can readily „bgjd^nJjfijOii^Jiy t^ They fall into two main types: (a) symmetric, and (b) asyia| metric. But this «mtřlu1it7^Gven~^ď7^^ dôéTTľot permit inferences about the ways in which allian<||| are actually contracted. A symmetrio terminology does no|| entail symmetric (reciprocals^ ~P^ern"Tritn^' ~~~~ ~~T~" ]naygovern^ä~^tricl^y^Íylmn^itri^transfc asymmetric terminology, on the other hand, is indeed incomf patible with symmetrio alliance, but nothing more positive! than this can be inferred. In systems with either symmetŕil or asymmetric terminologies, the exercise of preferences cafe lend a marked bias to the conduct of social relations, but'4j preference cannot necessarily be read off from a prescriptive terminology. m 4. The employment of any type of terminology as a practical]® instrument of social classification can be considerably affectedfl by innumerable factors which are not recognized in the termino|| logy itself. ■$$£ The mos^gejieralfjictor of the kind is relative age. Individuals! of the same age will belong naturally to different categories, and! individuals of different ages will belong to the same category!! Socjaj_rahxticji^jmay_therefore be ordered, in principle, eitherll by category or by relatjyelifflJľTf^^ criterion in classificationTtliěn in such instances it is to some|| extent incorrect to regard the terms as denoting distinct classes« of persons (Needham 1966a). ■'r'm Other common factors are residence, which Kroeber indeedÉ argued to be of fundamental importance (1938), genealogical« degree, and collective sentiments. There are in addition in-J| numerable further possible factors which affect the employment^ of terminologies but cannot be inferred from them. They can'il 20 M iJinJHtiiA'ti mt Im: ,1 urn !)■•(■> >;/ n in.,m j) iinu .i.m,,..,i, be discerned, and their consequences gauged, by the rical investigation of each particular case, limitation of a fundamental kind is placed on the interpre-i of relationship terminologies by the fact that we cannot i infer anything about the degree of significance that a ory may have, is was so well recognized by Murdock, in Social Structure, he went so far as to isolate what he called a criterion of ateriality'. By this he meant 'a negative similarity result-rom the functional unimportance of the relatives of two ppes, whereby a sufficient basis for differentiating them is ig' (1949: 136). Leach, similarly.in analysing the Jinghpaw m, stressed that 'On the one hand, individuals are classed her because, individually and as a group, they stand in a leant and important relationship to the speaker; but on the hand they may be classed together precisely because are unimportant and remote' (1961: 52). e latter case is well exemplified in systems, as among the in, of asymmetric prescriptive alliance. In these, the ories of lineal relatives are of the greatest jural importance, b is in accordance with this importance that more termino-il discriminations are made among these than among is. The terms for affines, wife-givers and wife-takers, com-a universal classification of social relations. There are three major classes of persons and groups: wife-givers, ; relatives, and wife-takers. Actual contact will be main-d with some people of all three classes, but how are those classified with whom there is no recognized relationship? solution adopted by the Haka Chin of Burma and the boru of Sumba is to classify them as agnates. So the terms seal relatives can apply both tojtj^_mo^important people e social universe and to the least jmportant. But there is eans of guessingTlnTIroňra^tudy of the terminology. In ■ words, we cannot even tell whether, or in what circum-es, a term will mean anything or not. of this looks pretty negative, I admit, but I should not it to be thought for a moment that I underrate the signi-3e of relationship terminologies. What I am arguing against , Éis, in the first place, the conventionaltypoTo^ý~^vhičTr^n5yTTiě / !■■■ 21 very lad of assigning terminologies to substantive types -le' I |u¥Tiolm|i^^ jflogy tells us more than it ever can. Secondly, I am argüi jl against the related assumption that societies witB~sTraB> f temnnologies, in the conventional typology, aré tfiiŕefi" f sl)c7cTÔlnc¥flTliírnu^^^^ do not aM mean that I am opposed to the analysis of terminologies or ev* to the construction of typologies. There are good and íl) rnetbofis_afjuiajysjs^jifteroM^anŮ^^^J^KS^J^^^^1 Cdation8hJp_jtejcmJDdogi^^ significance in the analysis and comparison of institutions an" s^cTaTäclblÖn (Needham ^Qg^rfQeff, egp^daJIly^Ônče^^realik what limitations we are m^e^Jn^m^jnterjgretaj^n^of theW The 7pTStÍôn"1s™\vEether there is some method, or conceptu revision, by means of which we can escape the prejudices an confusions into which received anthropological ideas tend;ti load us. The means to an improvement in analysis has, I think; already been provided by Lowie, basmg_jnms^lf_on_Kroeber,8 brilliant paper on classificj^Qjy^yjstems^fa^laiionship (Kroebér 1909). Instead of talking about a Hawaiian system, for example, and then being forced to decide whether a given case was or was not Hawaiian, he isolated a 'Hawaiian principle', i.e. a modo of classification which underlies non-lineal terminologies.". Instead of adopting the Dakota system as a paradigm, he abstracted the Dakota principle , i.e. that which orders hnea teHTmol^es. Then there were two~Variants' on the Dakota scheme: the Crow and Omaha principles of classification, i.e:' matri- and patri-lineal identifications. Finally, he concentrated on the terminological correlates of special forms of marriage,'? including 'prescribed' marriage with the bilateral cross-cousin,« a form which he saw as 'mirrored in the nomenclature' of certain* Melanesian societies (Lowie 1917: 107, 119, 150, 151, 152, 172).' Thus a terminology did not need to be labelled as a whole, and thence be assigned to one type or another, because of certain t typological features that it happened to possess, but was re-'S garded as 'molded by disparate principles'. This not only came | closer to the facts of the matter, but it also provided a clearer ** view of the historical change of terminologies; e.g. instead of ^ a clumsy succession of types, the Hawaiian principle could bo 29 JiUittti uö i-in Hu. í\ L llötl t) '!/' ífas gradually taking over from the Dakota principle, pro-ihg a variety of terminologies of which none might wholly respond to either the Dakota or the Hawaiian type, 'erhaps this will not sound like very much, but it was a great Tto take and I do not think that essentially we can do much iter; The central advantage^ofjtawie^sjnefji^ Ilysis it displmiei~wltjrj^ Slě¥TňstěayojaJrhe operation of classifica-tion. Instead of' lyin^fupon a paradigm case, and a class of other cases bearing j ^oŕe or less attenuated resemblances to it, the method investi-is the intricate combination of classificatory principles by 'filch any particular terminology may in fact be constitutedij e craving for generality is checked, and thejojie^ifiihiLessentiftl In the study of relationship terminologies as well, therefore, ^re^find that analytical advance is blocked by the familiar conceptual habits which have so much hindered the study of (descent systems. The remedy is not only urged by Wittgenstein ľut has actually been put into prior effect by Lowie. In_this rtSase too, however, there is thejjongequence that^ while analysis Unmade more exact, comparison is made_more_jntricate and difficult. In fact, it is hard to see how a large-scale comparison oaň~f5ě~cárried out when each case is analysed rather than typed. ^Nevertheless, it may be that this line of argument will be i|found far too simple, perhaps even obvious. Let me just quote, "therefore, what an authority (Murdock 1955: 361) has not long ago concluded about the study of descent systems and relationship terminologies: -i 'In anthropology, the initial classificatory task has now been substantially accomplished in the field of social structure. . . . We now possess satisfactory criteria for differentiating types of family organization, kin and local groups, and kinship terminology and behavior patterns. . . . These features are combined with one another in particular ways to produce a ' finite number of types of social organization, which in then-totality represent a systematic classification comparable to those of Linnaeus and Mendelyeev.' Well, I do not wish to disparage Murdock's decades of industrious application to these matters, but I am bound to say that 23 mni it! li jS CI'Mllltlll I think these statements are mistaken in every particular.^!? notion of jLÍjilŽM-ft"d total classification is logically indefensibl and this methodological^ ambition has achieved no result which might give it a pragmatic justiflcationu" ——-^ Yet theTdeET^continues to guide research. A recent illustr tion is furnished by Romney's analysis of the Kalmuk Mongl terminology (1907). It is a long and painstaldng piece of worl and I should not desire to call its analytical precision int question, but the formulation of its theoretical conclusions displays very clearly the typological concern which, especially since Radoliffe-Brown, has had such a deleterious effect 0$ anthropological thought. It is simply that the Kalmuk termini? logy cannot be accommodated by Lounsbury's rules and liencl that the differences 'warrant the recognition of an additional* typo of Omaha system' (Romney 1965: 141). Now this might be* a positive result if only it meant that the Kalmuk Mongol system^ could be assimilated to a class of social phenomena about whioh there was a body of tested theoretical propositions. But merely to] add a further type to the catalogue of 'Omaha' systems, whenj nothing enlightening or interesting has been said about this ill-a founded class of systems, does not seem a very useful exercise.^ m VI INOBST >' ''"i Lastly, I want to make some remarks on the concept of 'incest'.*. This is a notion that is as prominent in social anthropology, and) as persistent, as 'kinship' has been. The contributions to whafc^ is called incest theory have been very numerous, elaborate, j| and ambitious, culminating (as far as theoretical extremity is á concerned) with Lévi-Strauss's dictum: 'The incest prohibition | is . . . culture itself (1969: 12). $, The variety of explanations ju'oposed has been remarkable: M fear of menstrual blood, harmful genetic effects, instinctive % repugnance (Radcliffo-Brown even declared that 'It is this & emotional reaction that we have to explain if we are to have a Sa theory of incest'), disruption of the solidarity of the family, v disorganization of the status distinctions preserved by tho <; prohibition, and so on. But two general assumptions are shared ; 24 MH &ŤÍ . Memarks mi Utc Analysis oj Kins/tip and Jlarriaye m\ řithe: incest theorists, however much at odds they may other-"v»bo.- (1) That we know what 'incest' really is, whatever l"pits regulation may take. Thus Radcliffe-Brown easily irts, as though it were ta be taken for granted, that 'incest properly speaking the sin or crime of sexual intimacy between nmediate relatives within the family' (1950: 69; italics sup-iedp (2) That the prohibition of incest, in spite of its highly lied patterning, is a universal. Thus according to Lévi-Strauss ia-'a rule, but a rule which, alone among all social rules, aseases at the same time a universal character' (1969: 8-9). 'It is curious, therefore, that there is nevertheless nothing like ly general agreement about the explanation of incest prohi-iítíons. Yet the endless academic debate is carried on as if the ||anegated and often unrelated theories were (1) rival explanations of the same phenomenon, and hence (2) universally applicable. 'Disputes are multiplied, as if every thing was uncertain; these disputes are managed with the greatest warmth, as f every thing was certain' (Hume, Treatise, Bk. I). «foThese circumstances give rise to the suspicion that here too Ibmothing has gone radically wrong. The trouble may lie, namely, not in the comprehension of the facts or in the ingen-»uity of the explanations, but in the conception of that which is f'to be explained. It looks as though the classificatory concept |of 'incest' may itself actually have conduced to the confusion, ?»nd to the negative results, which characterize the present {theoretical situation. V Revel, in his irreverent critical comments on Les Structures *élémentaires de la parenie, makes this excellent and revealing 'observation: 'If there is one thing that emerges from Lévi-' Strauss's book, it is that it will not do to talk about the prohibition of incest in general, because the institutions falling under this concept are so diverse and sometimes contradictory' (1957: ^154). It sounds a simple enough point, but I think it contains 'the key. Incest prohibitions do not in fact compose a definite if class, and if this is so there cannot be a general theory that "'applies to all of them. There are two prime considerations which support this view. [<■ The first is the wide and variable range of statuses to which the 1 prohibitions ajiply. The scope of application is in each case an integral feature of the social system, and in some sense a 25 function of it; i.e. tlie complex of prohibitions in a sociéř cannot be comprehended except by a systematic purviews the institutions with which they are implicated. By this accolul of the matter there are as many different kinds of incest pn hibitions as there are discriminable social systems. -{š The second consideration is that incest prohibitions areji part moral injunctions; they are expressions of indigenoS ethical doctrines and, whether or not they are touched with!? peculiar emotional quality (something I doubt as a genera concomitant), they have cultural meanings which no attempt'! explanation can reasonably neglect. ř's«á Something of these latter connotations can be seen in thej disparity of the words which ethnographers translate by th|| English word 'incest'. Let us begin with this word itself, takinfi it not as a quasi-technical term in social anthropology but] philologically, as a particular testimony to a cultural traditiofff that remains implicit in its composition. It comes from the* Latin castum^ chaste, and bejojngsi^qj^JarailjofLBpjnanoéj words with the same root; e.g. French inceste, Italian incestofö Spanish incesto, etc. The idea ultimately^ímderginglliFffiglishí word™sc1HíŕšHthliiFóT^ offence againati piirítyandm^raHlecency. The Germanic family of languages,^ however, makes a significant contrast. In this we find related^ words such as German Blutschande, Dutch bloedschcmde^ Norwegian bladshande, etc. These terms are all compounded 013 ^f^^2!É^^^2^^^^^^^:::^^^^^n^e implicit^ idea, then, is not apparently about the guilt of sullied purity! but about the public disgrace of an offence against kinship;1« conceived (as we know from other evidence) as a community off blood. Next, to go far outside the European tradition, the f Chinese Jerm hum lun implies ^more^ocaological view. It is s composed of clmracters^meaning respectively 'disorder' and, 'soiciaTreTätTônllIip' Jäljsoj^jdnd^jck^ In cTäsšTcTíť^h1n^ó~poJitical theory the stabinty~ôTlne-štate depended on the 'five relationships' (wu lun), viz. sovereign/ ' subject, father/son, elder brother/younger brother, husband/* wife, friend/friend. The offence in this case, then, is that of disrupting a jural and moral order, namely that of confusing statuses. Finally, as a further culturally distinct example, let us take the Indonesian word sumbang, a word that is found in 26 iie'^ör another form (e.g. suvang) in probably the majority of frislands of the archipelago. Sumbang is commonly rendered ^'incestuous', but it also means 'improper' or 'repugnant'. ťSócial conduct it refers generally to what is offensive because jfefout of place or unseemly, a subversion of propriety. For-aderi sexual congress is only one form of such unsuitable jhaviour. For that matter, sumbang covers not only incest but gjgiQuple adultery as well. And cheating at cards or some other llpune is also sumbang. Very prominently, moreover, the word lllpläesthetic meanings. A connexion between the sexually KSpSiiig and the aesthetically shocking is made revealingly by a lipirase in horticulture: a tree which has been grafted so that it |||lars i blossoms of two different kinds is a pohok sumbang: llliroěptually, the word applies to what is deformed, dishar-llpoiiious, or discordant. A person's voice can be sumbang, hard Ipiuihe ear; and an everyday material object, such as a battered l||iiess-board, can also be sumbang, unsightly. Incest, therefore, ||jřiňdeed denoted by this word, but it would be a mistake for fwi?ethnographer to translate sumbang simply by the English Spur j ,. , , Ijeord incest. Ipln these four linguistic areas, etymological indications or current usages thus exhibit a variety of meanings which cannot validly be classed together as the semantic component of the ^prohibition of incest. ipílri addition to all this, there are cultures in which there is no £'8U'ch explicit notion as that of 'incest'. In fifth-century Athens, tíbt only was the verbal concept entirely lacking but there was mó action at law which was or could be brought against those ipiilty of an offence of the kind (Harrison 1968: 22). '#'By the jabove_jaríteria^^ prohibitions do not compose a class of homogeneous social Sphenomena. The most that might be claimed is that they ^characteristically include certain minimal prohibitions, e.g. |that on sexual intimacy between mother and son. This is in ffact disputable as a universal proposition, but even to concede |it would still not justify the conception that these supposedly ^'common features are what essentially define the class which is r'the subject of incest theory. It is this presumption, however, rwhieh suggests the received idea that the differential patterns of prohibition are 'extensions' of certain basic interdictions. 27 < -i*S iif« m m « ,"-;'í TIhih JMiirdook writes that 'incest taboos and exogamous resfcri, tions of whatever sort seem clearly to be extensions of the M taboos between parent and child and between brother and sis in the nuclear family' (1949: 284). This of course is neither logi; inference nor established fact, but the very presumption inY cates the diagnosis of the initial conceptual error about ince~ In this case also the difficulties arise from the invalid constiii tion of a class and, especially clearly, the unbounded exerc"' of tho craving for generality. £ We obviously need to find a radically different solution'*.., the problem of incest. A structural theory is unlikely to servo since there are no systematic resemblances among all kno~' sets of prohibitions. A formal approach via logical possibilities as in the reinterpretation of the concept of descent, does no seem feasible either, for there are no self-evident or readil acceptable premisses. Nor, finally, should we commit ourselv. to a reductionist type of theory, for this is precisely what mos anthropological explanations of the prohibition of incest hav been, and they have not worked. We have, it seems, to look J a quite different way of understanding the facts. Before we do so, let us make a brief appreciation of the; situation. Each set of cultural prohibitions forms a coherent but variablo assemblage of rules; the contextual explication off these rules demands a recourse to history, language, moral concepts, and many other contingent particulars. The various sets of prohibitions thus do not compose a concrete class such- as might be open to a unitary explanation; the functional or soman tio explication of the rules of one society may not apply;' at all to those of another. J Here, then, is a possible line of argument about incest theory," 1. In each instance, we are dealing with explicit rules, i.e. with collective representations. These rules may perhaps have emo| tional concomitants, and for all we know may even bo grounded ultimately in some common instinct or psychic character, but; empirically we can deal with them only as social facts. ;i 2. The rules, by definition, have to do with access to women.'[i 3. Women are social valuables; for many peoples, and in the? opinion of some anthropologists, they are the prime valuables!'!, 28 w Remarks on the Analysis o j Kinship and Jlarriage liAocess to socially recognized valuables is always socially (filiated; the regulation expresses the evaluation. |||Bules define what is permitted and what is prohibited. |§|The regulation of access to women is in these respects pjiti>ilike any other kind of regulation; access to certain categories of women is permitted, and to certain others is prohibited. These rules, positive and negative, compose in any J&rticular instance a coherent set and should be compre-fiended as such, i.e. correlatively. It is methodically nefeotive to consider oidy prohibitions, apart from permissions Mid^prescriptions, and it is yet more invalid to consider only Isolated prohibitions as though these alone were definitive or or*,-,. / b Essential. jf&What is at issue in the study of 'incest' prohibitions is simply Stejnegative aspect of the regulation of access to women. All iät is common to incest prohibitions is the feature of prohibition itself. |I conclude that 'incest' is a mistaken sociological concept/f \jj~t ~^^^x^vSíS^7r^^tQ~čä^^äJčä\& no general theory of f .,«/.*< liti pacešC «•z n^i HJt.***. t H CAS t " cu-c. p ppi- jfcSfre«ct. ^Bíifff SHUČÄ^ OONOLUS/ONS >1>Á*'&? % on .* hst in h cms _, <4v^^-*. i '% i/asté $ ilJiYk^Jl: |The intention behind these remarks has been to argue that anthropological research has been misguided, in even the supposedly basic topics of kinship and marriage, by certain con-icéptual failings. The most consequential of these is the uncritical |afctachment to an inappropriate conception of a 'class', namely Éthat a class is a number of objects possessing certain attributes, ! I//I in... . i Hutij^to tij líi/tů/iijj ana Jim ľuiijc SWhat is common to descent systems may be only that they yariously exploit certain logical possibilities in a sexually defined íálismission of rights from one generation to the next. What is immoa to the myriad forms of marriage may be only the con-aotual union of sexual statuses. What is common to relation-Op'terminologies is the exploitation of certain formal possi-Sitiea in the classification of sexual statuses. What is common jjpneest prohibitions is merely the fact of prohibition. There íftý'thus be something in common, under each general term, rat not necessarily a definite set of characteristic, specific, or ásential features. The common feature of prohibition, for »ample, does not entail that the incest regulations of different iííbieties shall in any specifio respect be at all comparable. ^Another tendency which Wittgenstein argues is responsible for'the craving for generality is our preoccupation with the Method of science. 'I mean the method of reducing the explanation of nattiral phenomena to the smallest number of primitive latural laws . . .' (1058: 18). In social anthropology this pre-focupation has most notoriously taken the form of a search for Siociological laws, and although this particular ambition is now iWlatively out of favour, or less grandiosely declared, the out-;I6ok still flourishes in the attempts to find statistical correlations. Now these cannot be established without a precisely ^specified typology of phenomena, and it is necessary for the Iptirpose that the phenomena classed under one type shall have yn common certain specific features. But if it is conceded that H'the social facts in question do not necessarily compose a con-flventional class of this homogeneous kind, but may exhibit instead an immense array of serial and more complex resemblances, then the grounds for this method of comparison and ^explanation are removed. |feThis does not mean, though, that comparative studies are unfeasible or should be put at a discount. On the contrary, if |we give up the reificatory typologies that are usual in social ránthropology we shall actually be in a better position to com-ifpare, for we shall at least see the social facts in a less distorted Sway. More positively still, we can carry out comparisons by preference to logical features, and by the formulation of more ^'suitable abstractions than have been customary. I have given Iff some indications, here and elsewhere, of the formal approach. W 31 I I lunt lift) i\ cc.tll/aiil That of improved abstraction may readily be illustrated by,ť study of prescriptive alliance. This typo of organization defined by the terminology, and the terminology is constitu by the regularity of a constant relation that articulates liri and categories. It is by reference to this relational abstraction not to lineages, groups, offices, and other such institutions" that effective comparisons can bo initiated. . <-* For that matter, I think that there are not only logi features to which Ave can resort, but that there are psyoW features as well which can be recognized through the screen^ cultural differentiation and which make comparison possibl But that is another matter; at any rate, it does not seem direotl relevant to the analysis of kinship and marriage. Yet it poiii in the same theoretical direction. If I may offer a more speoula tivo observation, it may be that all social anthropology will b" able to do - or porhajjs what it may best be able to do - is to comprehend, in one case after another, the schemes in whioh cultures have variously taken advantage of logical and psychi facilities which are the elementary resources available to mankind in the ordering of experience. ""■ In this case, to adopt a phrase of Wittgenstein's (1953, seo* 90), 'our investigation .. .is directed nottowmdaphenomena, but, as one might say, towards the "possibilities" of phenomena'. jb&/Ó^'^ &!&&*-' ACKNOWLEDGEMENTS ,*M.' 'l y t I am indebted to Dr Francis Korn for calculating the number of combinations. 'tym'.tfrtf &"*ť^*^" of the six olemontary modes of doacent. Drlon Hamnett haa kindly worked out? life}»' 4í/~ ft**.-- V^'^^—'othor posaibilitioa, on tlie basis of permutations instead of combinations, which J-fc-Sj! j! jL-i^- -> show that according to the assumptions and methods adopted the total could s Mi\P/,*SL" " J- bo 325, 1,056, 3,125, or 40,050 typea. . , l , j 'L/ ff f %!^3^ 4 tmi. *L4 K^s ->*ä*«5 O 4 ..-.-*, h REFERENCES . ■» ábfm-Of FOX, BOBIN. 19G7 KiluhinaJl M ^ ^t, M<~ Perspective. HarmonľsZrth MiddľslTpeťľ ^T?*** ŕV^iWot/^ w»»^ ■• a. (ed.). 1905. Forma SernantiľA ,' °°Y &lľ II , (Ĺ An^ropologisl ISveoia] PnlwTnVľ iL "A™1?**- -American f i* da^i Anthropological ACLln ^ **' ^"^ Ameri-" ■Kí >í£-6 ílísäH if iurC. at / / / řy-7- 4,-7 íu-e f-^c^AM-t'íf'C 32, &*u&~*>* 1 Remarks on tlie Analysis of Kinship and Jlarriaye IBIS on, a. B. w. 1968. The Law of Athens. Oxford: Clarendon Press. ART, A. M. 1952. The Life-giving Myth. Edited by Lord Raglan. (2nd impression, ed. with a>foreword by Rodney Needham, 1970.) London: Mothuen. 'E, david. 1739-40. A Treatise of Human Understanding. London. über, A. L. 1909. Classificatory Systems of Relationship. Journal of the Royal Anthropological Institute 39: 77-84. 1938. Basic and Secondary Patterns of Social Structure. Journal of the Royal Anthropological Institute 68: 299-309. OH, s. R. 1901. Rethinking Anthropology. (London School of Economics Monographs on Social Anthropology, No. 22.) London: Athlone Press. [-Strauss. 1969. The Elementary Structures of Kinship. Rev. sd. Edited by Rodney Needham, translated by J. H. Bell, J. R. von Stürmer, and R. Needham. Boston: Beacon Press; London: Eyro and Spottiswoode. IS, I. M. 1905. Problems in the Comparative Study of Unilineal Descent. In The Relevance of Models for Social Anthropology (ASA Monograph 1): 87-112. London: Tavistock Publications. IE, B. n. 1917. Culture and Ethnology. New York. dock, GEORGE feter. 1949. Social Structure. New York: Macmillan. 1955. Changing Emphases in Social Structure. Southwestern Journal of Anthropology 11: 361-70. DHAM, rodney. 1966a. Age, Category, and Descent. Bijdragen lot de Taal-, Land- en Volkenkunde 122: 1-33. 1960b. Terminology and Alliance, I: Garo, Manggarai. Sociologus 17: 141-57. 1967. Terminology and Alliance, II: Mapuche, Conclusions. Sociologus 18: 39-53. 1969. Gurage Social Classification: Formal Notes on an Unusual System. Africa 39: 153-66. lee, L. 1949. Naar Inleiding van de Stuwdam in Mangili. (Opmerkingen over de Sociale Structuur van Oost-Soemba.) Bijdragen tot de Taal-, Land- en Volkenkunde 105: 445-59. OLiFFE-BROWN, a. r. 1950. Introduction (pp. 1-85) to African Systems of Kinship and Marriage, ed. by A. It. Radcli (Tc-H» Brown and Daryll Fordo. London: Oxford University Press for Ifi' the International African Institute. ľ 33 "8a !! 'i &* Rodney Needham HEVEL, JEAN-FRANgois. 1957. Pourquoi des philosophes. Paris Pauvcrt. noMNEY, a. kimball. 1967. Kalmuk Mongol and the Classification / of Lineal Kinship Terminologies. In Hammel (ed.) 1967:127-41, Wittgenstein, Ludwig. 1953. Philosophical Investigations, Translated by G. E. M. Anscombe. Oxford: Blackwell. ---- 1958. The Blue and Brown Books. Oxford: Blackwell. © Itodnoy Neodham 1971 -I 34 Mi