Model and Metaphor in Social AnthropologyD. H, ReaderIt is a truism that as social scientists we areconstrained in our thinking by the dominantbeliefs and intellectual climate of our time. Kaplan(1964) has pointed out that some values in scienti-fic beliefs arise through what he calls key meta-phors in various periods of history: fruitfulcomparisons which stimulate the imagination ofthe age. The eighteenth century was given toclockwork conceptions, the nineteenth to organis-mic ideas; and now in the twentieth century wetend to make scientific formulations in terms ofthe workings of a computer.What is noticeable about these conceptualdevices is that (a) semantically they are all meta-phors, (b) functionally they all serve to relateevents or discoveries into coherent systems,(c) formally they are in some sense models ofreality for heuristic purposes, and (d) they allinvolve the implied use of analogy.The intention in this paper is to review the useof model and metaphor in one social science,social anthropology; to come in this context to abetter understanding of the working distinctionbetween these terms; and finally to appraise theirusefulness and comprehensiveness for thisdiscipline.EARLY MODELSThe model in some sense has been at leastimplicitly used since the beginning of socialanthropology. It was originally connected withsocial change, a subject in which many workers ofthe nineteenth and early twentieth centuries wereinterested. Against a background of Darwinianevolutionism and contemporary anthropologicaland historical studies (e.g. Bastian 1860; Buckle1857-61; Lubbock 1855; Waitz 1858) they assumedthat a given society was both stable and at acertain stage of development. The problem wasthen to discover or surmize what had happened toit before this stage, and what further changes itwas likely to experience in the future.In general terms, all societies were posited aspassing in social evolution through the stages ofsavagery and barbarism to civilization. This wasan organismic model into which various societiesexamined (usually from the literature) could befitted and classified. The apotheosis of thisscheme was of course Victorian society from whichthe investigators themselves sprang and had theirbeing. Nevertheless, it had all the rough-and-ready taxonomic advantages of an early model.Moreover, as a good model surely should, itstimulated research. Using this model, Morgan(1878) and McLennan (1897) devised an evolution-ary sequence for the family. Correspondingsequences for law were suggested by Maine (1897)in England and Letourneau (1890) in France. In thesphere of religion, Tylor (1865) and Frazer (1890)postulated appropriate stages, as did the Frenchworkers Lefevre (1891) and Mauss (1896).Many of these writers accepted foreign andbizarre customs as evidence for the validity ofpostulated but unobserved stages of the model.Where fact would not support fiction, the modelwas deduced by argument. Paternity, it wasmaintained, could never have arisen until the stageof polygamy was reached, Hence patrilineal de-21scent and inheritance necessarily belonged tothe higher forms of society (Achelis 1890).Models of this kind, with only one central themeand few detailed ramifications, might be calledunithematlc. They seem to have an essentialplace in the early stages of enquiry when the fieldneeds to be delineated in broad black and whiteterms. As more sophisticated material comes in,they become more and more naive and unsatis-factory, and, their job done, are finally repudiated.As the evolutionary unithematic model wasgradually abandoned in the face of mountingscientific criticism, its place was taken for a whileby a diffusion model. The problem here was totrace the spread of culture items, mainly techni-ques and material objects, through past migra-tions. Apparently, however, the time was not yetripe for inductive verification by fieldwork, althoughthe model made this possible and desirable.Perry (1927) tried very persistently to show fromthe literature that wherever practices such asbuilding pyramids, mummifying the dead andwearing gold and pearl ornaments existed, theyhad been introduced by the ancient Egyptians, orby people in contact with them. This model muststill be described as unithematic.Hogbin (1958), to whom much of the material forthis summary is due, believes that in England amodern reorientation set in just before 1920. Theyounger anthropologists realised that the methodof conjectural history, as Radcliffe-Brown latercalled it (1952), was unscientific because theconclusions could never be verified. At the sametime it came to be generally accepted that studentsshould not only visit the areas of investigationthemselves, but should learn the vernacular andengage in participant observation. Yet the in-fluence of the earlier evolutionary model and itsoperators was felt for another decade, at least tothe extent that students continued to be attractedby the unfamiliar. They selected remote societiesfor study, and where they were obliged to investi-gate a culture which had "broken down", theytried through the tribal elders to recapture theunspoilt past. Malinowski himself, to whom somuch of modern fieldwork technique is owed,makes only casual references to the Europeansand missionaries who had been living in theTrobriand Islands for fully twenty years before hefirst reached them in 1915 (Malinowski 1922; 1935).Still another characteristic was involved in thenew mode! to come. As Mitchell has indicated(1961), Malinowski, and Radcliffe-Brown in hisearlier period (1922), studied small-scale islandcommunities in which it was possible to see thewhole society in action. Redfield denoted this asthe "period of study of simple societies con-ceived as self-contained autonomous societies"(1955). Whether such a limitation of field helped todetermine the functional biological model or wasin turn determined by it is difficult to say at thisdistance. In any event, the new model transfixedin immobility societies which had previously beenseen in process of change. The model requiredthat the parts of society should be seen in func-tional interdependence, after the analogy of theneed-satisfying components of a biologicalorganism. It was hard enough to analyse out thesocial parts in interrelationship without the addeddifficulty of treating these parts as variables insocial change. The institutions of society wereaccordingly taken as constants in terms of theneeds they met. This connected with a furtherpresuppositionŠimportant for later workŠthattheir total interactional condition was one ofstable equilibrium. It is significant that Malinow-ski's later attempts on the basis of this model tointerpret social change and "culture contact" aregenerally acknowledged to be the least successfulof his work (Gluckman 1948; Malinowski 1945;Mitchell 1961), On the other hand, it was obviousfrom his writings (e.g. 1926), that Malinowski wasfully emancipated from the errors of "conjectureor hypothetical reconstruction" and was anxiousthat his model of society should correspond withreality. This is the first anthropological modelwhich could be called truly polythematic.SOCIAL STRUCTURE MODELSIt fell mainly to Radcliffe-Brown, influenced nodoubt by his philosophical and logical training atCambridge, to develop the functional biologicalmodel into a biological structural one. Functionalinterdependence on the analogy between socialand organic life was retained (1952) and conscious-ly referred back to its early formulation in Durk-heim (1919). While the dangers of the biologicalanalogy were fully realised, it was extended in anot particularly helpful way to designate areas ofstudy as social morphology, social physiology andsocial evolution (Radcliffe-Brown 1952). Theconcept of general human needs taken up byMalinowski was played down by Radcliffe-Brown,and replaced by a narrower and more technicaldefinition in terms of "necessary conditions ofexistence".This was required because of the notion of"structure" which Radcliffe-Brown superimposedon the biological model. Even in Malinowski'susage, the social organism was a closed structureor system, in the sense that it had outer physicallimits and was internally self-functioning. More-over, the term "structure" was already in currentuse, often in reference to the form of persistentsocial groups. But it was Radcliffe-Brown who insocial anthropology gave ordered expression tothe concept and embodied it in his functionalworking hypothesis. Using a metaphor, he definedsocial structure as a "network of actually existingsocial relations", but included also the differentia-tion of individuals and classes by social role. Thisstructural emphasis led attention away from theorganismic aspects of the model, which in effectbecame metaphorical in character. Human needswere less directly important than the harmoniousfunctioning together of the parts of the socialstructure. Function was now defined in terms of"sufficient . . , internal consistency, i.e. withoutproducing conflicts which can neither be resolvednot regulated" (1952: p. 181).While the assumption of general human needscould be put aside in this scheme, Malinowski'sconcept of stable equilibrium was as important asever. In terms of the organic analogy, Radcliffe-Brown considered deviations from stability aspathological. Disequilibrium was given a valuejudgment as a diseased condition of the socialorganism, producing dysnomia, disorder, socialill-health, as opposed to the eunomia of thehealthy integrated society (1952: p. 182). This valueassumption was to have unfortunate conse-quences, not only for the effective study of socialchange, but for later investigations of large-scale,complex societies,Homans was an early critic of the biologicalstructural model. He found fault with the argumentthat because some recurrent activity is "organic-ally" interrelated with other activities, it neces-sarily makes a contribution to social survival,"The interrelatedness of the elements of socialbehaviour may be dysfunctional as well as func-tional", he wrote (Homans 1951: p. 271). Hogbin,too, has questioned Radcliffe-Brown's organicreiflcation of society: "There is no point in sayingthat a certain custom has an integrative functionfor the society as a whole; and to state that societyin a state of dysnomia always struggles towardseunomia is to cloud the issues. It is rather indivi-dual persons who combine into new groups whenthey find that some common aim is best achievedthereby ..." (Hogbin 1958: p. 29).The idea is tempting that over the years thebiological aspect of the biological structural modelhas dwindled from analogy through metaphor intomeaninglessness. What is clear is that (a) oncethe tendency to functional interdependence of atleast clusters of social institutions was accepted,the organismic analogy had fulfilled its purpose,and (b) the social structure model is capable ofseparate logically self-consistent existence. Theresult has been that various forms of socialstructure models have made their appearancewithout overt biological implications. Even these,however, have rested on the presuppositions ofclosure and stability characteristic of their biologi-cal prototypes.Leach has criticized these presuppositions, andin the process has produced another type ofstructural model (1954). Anthropologists, he says,are almost alone in regarding social change asshattering and somehow fundamentally immoral.Their prejudice in favour of integration, functionalconsistency and structural equilibrium is the out-come of the conditions under which they work. Ananalysis torn out of time and space in the courseof a year or two's fieldwork requires the axiom ofequilibrium, for without it the model would appearto be incomplete. But they go too far in assumingthat the equilibrium is stable, i.e. firmly estab-lished and unlikely to suffer sudden change.When the anthropologist attempts to describea social system, Leach goes on, he necessarilydescribes only a model of the social reality. Themodel represents the anthropologist's hypothesisof "how the social system works", and is thereforea coherent whole. This does not imply, however,that social reality is a coherent whole: on thecontrary it is full of inconsistencies, which providean understanding of social change. But in practicalfieldwork situations the material observed must betreated as /fit were part of an equilibrium system,or description would be "almost impossible".All that Leach asks is that the fictitious nature ofthis equilibrium be fully recognised. His descrip-tion of a social system provides an idealized23model which states the "correct" status relationsexisting between groups within the total socialsystem, and between the social persons who makeup the particular groups.The logical attempts made here to reconcile themodel with reality are interesting, and may besummarized under the heading of dissociation. Incontrast with Radcliffe-Brown, the model is firstlyof idealized status relations as opposed to actual"person-to-person" ones. It is an idealized struc-tural model, which allows unwanted idiosyncraticvariations to be dissociated or excluded. Again,the equilibrium of the model is dissociated fromthe disequilibrium of reality, and the two areconnected only by a tenuous "as-if" relationship.The model marks an advance in (a) dispensingwith the biological analogy which had becomeonly a metaphor, (b) recognizing fully the non-integrated nature of more complex societies, and(c) opening the door to an analysis of socialchange. The price paid is the new problem of thecriteria by which the anthropologist's model,since it is dissociated from reality, shall be testedagainst it.Anthropologists are not alone in being casti-gated for regarding social systems as closed andsocial change as shattering. Dahrendorf (1958)has taken the sociologists to task for much thesame fault. He has, he says, yet to see a problemfor the explanation of which the assumption of aunified value system is necessary, or a testableprediction which follows from this assumption.One of the more unfortunate connotations of theword "system" is its closure. Although somestructural functionalists have tried to, there is nogetting away from the fact that a systemŠif onlyfor "purpose of analysis"Šis self-sufficient,internally consistent, and closed to the outside.By no feat of the imagination can the integratedand equilibrated social system be made to produceserious and patterned conflicts in its structure.What it does produce is the well-known villain ofthe piece, the deviant. He quotes early Parsons:"Deviance is a motivated tendency for an actor tobehave in contravention of one or more institu-tionalized normative patterns" (1951; p. 250); anddeviance occurs "either if an individual happens tobe pathological, or if, from whatever source, adisturbance is introduced into the system" (p, 252).Dahrendorf recommends as a corrective aproblem-conscious discipline (1958: p, 124).Problems require explanation; explanations re-quire assumptions or models, and hypothesesderived from such models. These hypotheses arealways, by implication, predictions as well asexplanatory propositions, and require testing byfurther facts. Testing in turn often generates newproblems. He also emphasizes the usefulness of aconflict model of society, based on the presuppo-sitions that (a) continuous social change is takingplace unless some force intervenes to arrest it,(b) social conflict is ubiquitous: its absence issurprising and abnormal, and (c) societies areheld together not by consensus but by constraint.Dahrendorf does not insist that the conflict modelis the only one. Problems for study can be selectedin terms of the equilibrium model or the conflictone, or perhaps of other models too. He pointsout, though, that the models with which we work,apart from being useful tools, determine to nosmall extent our general perspectives, our selec-tion of problems and the emphasis of our explana-tions. He might also have indicated that thesefactors help to determine our models.In social anthropology, Gluckman and hisco-workers, generalizing initially on rural field-work in Central Africa, have for long urged thatconflict, ambiguity and inconsistency are charac-teristic of social change. Gluckman has empha-sized that conflicts in men's allegiances in one setof relationships lead to cohesion through cross-cutting alliances over a wider range of relations,orthrough a longerperiod of time (Gluckman 1955).Mitchell has crystallized the concept of conflict inthe notion of the "plural society" (1961), afterFurnivall (1948). Here, disparate systems ofcustoms and beliefs coexist, and are called intoaction in different social situations. Mitchell doesnot entirely accept Gluckman's theory of "counter-balancing cleavages", but seems more to agreewith Dahrendorf that constraint rather thanconsensus is the basis of cohesion in pluralsocieties. His eventual working model of "complexreticulations of social relationships" does not,however, appear to differ greatly from Epstein's"different sets of social relations or spheres ofsocial interaction" (1958: p. xvii). Gluckman hasexpressed a similar view of "loose, semi-indepen-dent, to some extent isolated sub-systems"(1961: p, 80). With all these scholars, despite theirfull recognition of conflict and inconsistency, it isnot difficult to discern the underlying search for a24systematic model: the "cross-cutting alliances"of Gluckman, the "complex reticulations" ofMitchell, and Epstein's "spheres of social inter-action". It would appear that the social structuremodel has been forced into "semi-independentsub-systems" mainly because of the disconnectednature of the data themselves.There are other versions of the social structuremodel which by their formulation avoid, or at leastaccept more naturally, the pressures imposed bysocial disunity. One of the most valuable is thatof Nadel (1957), who, following Parsons (1954),defines the structure of a society as the pattern ornetwork (or "system") of relationships between"actors in their capacity of playing roles relative toone another" (Nadel 1957: p. 12). The notion of rolehas been curiously neglected as a formal conceptin anthropological models (Reader 1961: p. 212).It enables Nadel to make the important distinctionbetween corporate groups, in which membershiphas all the characteristics of role-performance,and institutionalized social relationships, analysisof which proceeds pari passu (1957: p. 60). Occa-sionally, says Nadel, anthropologists have chosena mode of presentation whereby they single outparticular roles and outline social structure onthis basis, reaching it by way of a role inventory(e.g. certain kinship roles in Eggan 1950; Fortes1949), Generally, however, roles are only describedas they become relevant.The great advantage of the role structure modelis that it provides effortlessly for conflict, ambiguityand inconsistency without further presuppositionssuch as "as-if" relations or plural societies.Nobody has any difficulty in visualizing incompat-ible roles performed by one and the same individualor group; roles which are ambiguous, eitherdeliberately in a fluid social situation or becausethey are new and have not had time to crystallize;or roles which are inconsistent in one socialsituation compared with another. Moreover, Nadelmakes full allowance for cleavages in the structure.Absence of a common logical locus, he says,precludes the assumption of a unitary, coherentrole system. Indeed, there seem to be as manyseparate systems as there are logical role-frames.Between them there is only the linkage provided byrecruitment rules, defining the flow or "circula-tion" of persons between disparate sets of roles,and the chances of their belonging to several atonce (Nadel 1957: p. 97).Nadel is also able to deal with the principle ofequilibrium. Purely objectively, and ignoring theviewpoint of welfare, he says, human communitiesin a state of equilibrium are neither better norworse, neither more nor less interesting, than anyother state. The assumption of equilibrium isimportant only in that it makes sense to look fordeterminacy; and the constancies observed infield anthropology are certainly of short range(1957: p. 145). It is clear, however, that the short-term constancy of a particular logical role frame,not changing its shape "as soon as our backs areturned", is all that is required to satisfy the condi-tions of Nadel's model. If no workable constanciesemerge, then there is no society to study and nostructure to define.Borrowing from Levi-Strauss (1953: p. 528)Nadel calls social structure a "statistical model",in the sense that it has the same degree of reality(or, as he says, non-reality) which would beascribed to any purely statistical picture of a socialsituation. By contrast there are "mechanicalmodels" of societies, exemplified by their validlaws: marriage laws, for example "calling foractual groupings of the individuals according toclan and kin". It is only the pragmatic design ofsocieties, their body of rules backed by sanctions,which can be ascribed concrete efficacy and"real" consequences. The statistical model canhave no such effectiveness, but provides only"thresholds" (Levi-Strauss), zones of indeter-minacy, and hence indices of the probabilitywhich its Constances apply (Nadel 1957: pp. 147-148),This attempt to operate an idealized structuralform through the medium of an actual structuralmodel resting on the tradition of Malinowski andRadcliffe-Brown, leads to severe difficulties.Radcliffe-Brown believed he knew what he wantedas the basis of social structureŠthe "concretereality" of actually existing social relations (1952:p. 192). Gluckman, in an intermediate position,builds his model on "typical" or representativeevents (1942: p. 245). These may not accuratelydescribe any actual social events, but representthe type of behaviour that underlies actual eventsin a given community. This formulation shows akeener appreciation of the inductive first-degree-abstraction process by which field data areclassified from the level of observation. Finally,the possibility arises, as utilized by Leach, of25stating the "correct" or idealized relations whichthe anthropologist thinks should exist betweensocial persons.At this point it may be useful to make a distinc-tion between ideal, idealized and ideationalmodels. An ideal model is built on what aresometimes called normative patterns of behaviour.These are at the level of descriptions by infor-mants, or comments by them, on what "ideal" inthe sense of perfect or rational behaviour shouldbe. In the courts of law such behaviour is oftenlaid down, and actual behaviour assessed againstit, in terms of the concept defined by Gluckman inthe "reasonable (and upright) man" (1955b: p. 22),In an idealized model, on the other hand, "theconcepts and relations chosen ... are not given bynature, but are largely the invention of the investi-gator. They are governed primarily by the way inwhich he thinks it 'profitable to represent ex-perience' " (Firth 1954: p. 7). Such, according toFirth, is the economist's model. Levi-Strauss, too,points out that "social structure has nothing to dowith empirical reality but with models built afterit"; and these must be such that "they makeimmediately intelligible all the observed facts"(1953: p. 525). Leach, similarly, seems to think thatstructural analysis is only tenuously connectedwith reality. The structures which the anthro-pologist describes, he says, are nothing more realthan "models which exist as logical constructionsin his mind." (Leach 1954: pp. 5, 9).Nadel makes an attempt to square the idealizedmodels of Leach and Levi-Strauss with his owninclinations, moulded by the British school ofanthropology, towards an empirical model. Allempirical models, by definition, are inductive: thatis to say, their terms and relations are arrived atby the logical process of generalization fromrepeated, similar, particular instances of observedphenomena. This abstraction, at the first level, isperformed in the very act of recognizing andseparating out a particular phenomenon fordescription (Beattie 1959: p. 48). We thus neveranalyse reality directly, and to think that we do isto misunderstand fundamentally the process ofinduction. The "concrete reality of social struc-ture" is a logical contradiction: the basic terms ofany model, social structural or otherwise, are atleast first-level abstractions. Empirical models aretherefore at least inductive, second-level generali-zations of first-level abstractions derived fromobservation. So are ideal models, but they arebased on normative conceptions as framed byinformants. Ideational models, as will presentlyemerge, may or may not be inductive. Idealizedones are by definition not inductive but deductive.They are convenient, imaginative constructions,deduced from general experience to fit "reality"as seen by the operator,Nadel seems both to have misconceived induc-tion and to have been torn between empirical andidealized models. Following Radcliffe-Brown (1952:p, 192), he writes: "I consider social structure, ofwhatever degree of refinement, to be still the socialreality itself, or an aspect of it, not the logic behindit; and I consider structural analysis to be no morethan a descriptive method, however sophisticated,not a piece of explanation." (Nadel 1957: pp. 150-151). He admits, after Braithwaite (1953: pp. 90-91,108) that a model implies more than this, namely a"picture" so constructed that it has a logicalnecessity and explanatory power, so that verifiabledeductions can be made from it. But he does notthink that "social structure" satisfies this morerigorous condition. In point of fact, if the presentanalysis is correct, Nadel could have foundsatisfaction in either of two broad types of model:the empirical one, generalized from reality andmodelled on it, or the idealized one, deduced fromgeneral social experience and capable of beingverified by reference to it. Instead, his positionforced him to the regrettable conclusion that whatmakes structural analysis informative is not thefinal picture at all, but the analytic steps that leadup to it (1957: p. 154).To complete this analysis, there remains to beconsidered the ideational modelŠthe model ofsocial ideas. A classical exposition of this type isin the social philosophy of Emile Durkheim (trans.1953). Sociology, he says, studies a normativemodel of society, and not the distorted modelconstructed from an expression of public opinion(1953: p. xvi). There is an enormous gap betweenthe way in which values are actually estimated bythe ordinary individual and the objective scale ofhuman values which should in principle governour judgments. The average moral conscience ismediocre; it feels only slightly the commonestduties, and is blind to some (1953: p. 83). Moreover,because a certain condition is found in a largenumber of people, it is not for that reason "ob-jective". The "general" phenomenon itself, in26relation to the behaviour of the majority of indivi-duals, may be of relatively infrequent occurrence(p. xii). Such a conception of social fact rests onthe assumption that society is a system of ideas,beliefs and sentiments, working through, but overand above, the individual minds and ideas fallingwithin it (p. 59).This ideational model deals best, as one mightexpect, with the ideological and value aspects ofsociety. Civilization, Durkheim points out, is theresult of the co-operation of men in associationthrough successive generations. It is essentiallya social product: a "congregation" of the highesthuman values. We can receive from this store-house of intellectual and moral riches at most afew fragments. The more we advance in time, themore complex and immense does our civilizationbecome. Consequently, the more does it transcendthe individual consciousness, and the smaller doesthe individual feel in relation to it (Durkheim 1953:p. 54). Again, Durkheim contends that Socratesexpressed more clearly than his judges the moralitysuited to his time. It would be easy to show that asa result of the transformation of the old societybased on the gens, and the consequent disturb-ance of religious beliefs, a new morality andreligious faith had become necessary in Athens.This was not felt by Socrates alone, but also in apowerful current of opinion represented by theSophists. It was in this sense that Socrates wasahead of his time, presaging social change, whileat the same time expressing the spirit of the age(1953: pp. 64-65). Durkheim's analysis of "momentsof collective ferment", at which are born the greatideals upon which civilizations rest, is alsoilluminating (pp. 91-92).On the negative side, severe distortions andassumptions have to be made to accommodateindividuals and their relationships to the ideationalmodel. In order to fit the scheme, induced be-haviours have to be converted to ideas in the formof "social facts", which are then reified. When, forexample, one examines not individual suicides, butthe rate of suicide, "this total is not simply a sum ofindependent units, a collective total, but is itself anew fact so; generis, with ... its own nature , . ,dominantly social" (Durkheim 1953: p. 46). Thisapproach did not, however, prevent Durkheim fromapproaching the problem of suicide with a statisti-cal treatment much in advance of his time.Even ideas themselves cannot be subsumedunder the model at the individual level, for"society" is something over and above ihem,Collective representations, produced by actionand reaction between individual minds, areaccordingly created, on a fallacious analogy withthe supposed emergence of individual repre-sentations (ideas) from the interaction of neuralelements in the brain (1953: p. 27). Individualminds are not, so far as we know, like neuralelements. Neither is the connexion betweenindividual minds and collective representations(if these exist) at all demonstrably like that betweenneural elements and individual representations.The "emergence" of collective representationsitself is a metaphor, again highly dubious, basedon chemical combination, in which the product hasproperties over and above its interactive elements.Such conceptual devices are required to fulfil atleast three functions in Durkheim's scheme:(1) to maintain all elements of the model at thesame ideational logical level, so that it can bestructured; (2) to explain the normative force ofsociety admittedly felt by its members; (3) toseparate sociology from individual psychology.Thus Durkheim, in his model, is forced to reifythe social facts and collective representations ofsociety. He holds that social facts should beapproached as if they had a reality independent ofthe observer. Collective representations provide aconceptual framework of action (1953: p. xxii).Society as thus constituted provides "an estab-lished classification" of values, outside individuals,"which is not their own work, which expressesother than their own personal sentiments, and towhich they are bound to conform" (p. 84). AlthoughDurkheim elsewhere makes it clear that individualchoice is possible within and even against theconceptual framework of collective representa-tions (e.g. p. 61), a constant effort of will is neededso as not to confuse the emergent structure ofsociety with "a monstrous Group Mind" (p. xxiii).Functionalism as previously considered is notnecessary to Durkheim's model, which is accord-ingly freed from the social immobility inherent infunctional models. As Peristiany points out,"Durkheimian society does not balance 'as in abudget', ends achieved with energy spent. Usstandard of value is not that of the happiness ofthe greater number or of the average citizen. It isnot social utility or even the survival of societyunder its material form." (Durkheim 1953: p, xxviii).27Durkheim himself declares that to explain thefunction of ideas by the contribution they make tothe maintenance of the equilibrium, the solidarityor the survival of a society is to misconstrue thecentral tenet of his sociology, that individuals aresubordinated to society as a system of ideasDEDUCTIVEMODEL THEORYThe survey of structural and pre-structuralmodel-types which has just been made maybroadly be summarized in a table. This does notpresume to show the chronological or otherinfluence of any one form of model upon another:DEDUCTIVE/INDUCTIVEINDUCTIVE(Empirical)UNiTHEMATICPOLYTHEMATICEvolutionary(1855-1897)Diffusion(e.g. Perry 1927)Diffusion(e.g. Rivers 1914)Ideational(Durkheim 1919)Functional-Biological(Malinowski 1922-1945)Biological-Structural(Radcliffe-Brown1939-1952)Idealized(Leach 1954)Conflict(Gluckman, Mitchell,Epstein 1942-1961)Ideal(Gluckman 1955)Role-Structure(Nadel 1957)N.B.ŠThe above dates are only a rough guide in terms of significant publications.(p. xxviii). This both explains the persistence ofcustoms, as when individuals change their socialenvironment but cling to their ideals as symbols ofsocial identity, and leaves the way open for socialchange in terms of new ideas.It must finally be observed that ideational modelscan logically be either inductive or deductive inform. Durkheim offers a composite model in whichbehaviour may first be induced in terms of statis-tical rates. The result, however, becomes a unit ona new sui generis supra-individual level, and isconnected deductively with other units on thestrength of generalizations derived from analogyand metaphor. Moreover, the relationship between"social facts" as thus produced and "collectiverepresentations", whose mode of inference is notclear, is not fully worked out. A simpler ideationalmodel might well be evolved deductively, as whenone postulates from general experience what arethe operative ideas of a society; or inductively, aswhen they are inferred from questioning repre-sentative samples of social populations.It may be proposed, however, that increasingknowledge of society has so far tended to bringabout a change from deductive to inductive modelswith time. At the beginning, in the absence ofempirical knowledge, an idealized scheme has tobe superimposed on the limited data available,which their sparsity is often unable to contradict.As empirical knowledge grows, inductive proce-dures tend to shape the model more and more, forotherwise it will not cover, and is not verified by,field data. The second and more sophisticatedreturn to deduction in social science is suggestedby Leach (and by Firth and the economists). Theidealized deductive model is not in fact "profit-able" as a representation of reality until the opera-tor already possesses a profound inductiveknowledge by experience of the society he isinvestigating. It is, as Firth says, "a deliberateconstruct, simplified from, or departing from, reallife situations for heuristic purposes" (1954: p. 6).It has been described in the more developedscience of economics as "a closed symbolicrepresentation of the interaction of certain econo-28mic phenomena" (Stone and Jackson 1946;p. 555).From the various considerations adduced in thepresent survey of models, it would appear that asocial model in any form may serve the followingpurposes:1. To make description and visualization ofreality, or of real problems, more plain;2. To provide general interrelation or explana-tion of as wide a range of facts as possiblein terms of the minimal number of inter-connected general propositions or symbolsof them, resting on the simplest, fewest, andmost fully acceptable axioms;3. To indicate fruitful possibilities for furthertheorizing, problem-finding and experimen-tation within the field covered by the model:in a word, to promote further hypotheses;4. To predict, as in applied anthropology, whatwill be the outcome of certain combinationsof circumstances or variables which arecomprised by the model.These purposes evidently do not differ fromthose of social theory at large. But the contentionis that models are theory making use of a particularlogical form: analogy.ANALOGYIt seems logically necessary that all modelsbearing any reference to reality should be based onanalogy. This is indeed a type of inference bearinga strong formal resemblance to induction itself.The induction used in social science is nearlyalways of the kind called by logicians "inductionby simple enumeration" (Stebbing 1930: pp. 246-249). It involves counting a number of instancesrecognized as having certain properties in com-mon, and proceeding by inference to the assump-tion that all instances have these properties incommon. This is to be contrasted on the one handwith "perfect" induction (Aristotle) in which everyinstance is known, and with "intuitive" inductionon the other, the immediate apprehension of aninductive generalization on the strength of oneobserved instance (Stebbing 1930: pp. 243-244).Analogy, likewise, involves inference fromresemblances. But whereas with simple enumera-tion the inference is from similarity of the known tothe unknown within the same total class ofphenomena, in analogy it is between differentclasses which are alleged or believed to resembleone another in the properties and relationships atissue. To convert allegation into justified belief,and thus demonstrate analogy, it is necessary toshow what Keynes calls a strong "positiveanalogy" and a weak "negative analogy" (1921)between the properties of members of the originaland of the allegedly similar class.*ln other words,the respects in which the properties and relation-ships of the two classes resemble one anothershould greatly outweigh the respects in whichthey differ, these similarities and differencesshould be explicitly stated, and the differencesshould be irrelevant to the comparison. Sinceknowledge of the properties and relationships ofthe classes is never complete, the total analogywill always include properties and relationshipswhich are not known to belong, or not to belong,to the two classes being compared (Stebbing1930: p. 250).The suggestion now is that models are aconstructed class, or a set of interrelated conrstructed classes, with which the properties andrelationships of corresponding classes of pheno-mena seen "in real life" are being compared byanalogy. The constructed class may be indicatedby "expressive" signs, i.e. words, or by "substi-tute" signs, i.e. symbols (Stout in Stebbing 1930:p. 115). Examples in the latter case are the deduc-tive symbolic models of the economists andapplied mathematicians. These are symbolicrepresentations of the expected relationshipsbetween classes of phenomena: that is, syntheticgeneral statements by analogy of what the rela-tionships between corresponding phenomena inreal life can be expected to be. They may be grosslysimplified or abstracted from reality, and they maybe problem-oriented to a degree that involvesdistortion of reality as a whole. But unless theybear some analogic resemblance to reality, theyare surely not models of reality at all; and yetthey cannot be "models" of anything else. Withoutsome correspondence with reality, they are uselessfor that visualization, explanation and predictionof realityfwhich is their raison d'etre. Such modelsmay be termed symbolic models.Before proceeding to models expressed inwords, the useful representational model, notdiscussed in terms of symbolism by Stout, arises.This is probably closest in conception to theŁOther considerations, such as "comprehensiveness" (Keynes}, are left out forsimplicity of presentation."I" For philosophers who hold an "ideal "theory ot perception, it may be acceptablethat "reality" and the "constructed class" are mainly different sets ot sense-data between which a relationship of correspondence holds.29original use of the term "model": it is the physicalmodel with which we are usually first acquainted.In a typical model of the brain, for example, suchas is used for anatomical demonstration, thecerebrum in relation to the cerebellum, themesencephalon, the pons and the medullaoblongata are shown by physical analogy betweenmodel and original. The model is a medial formrepresenting many original dissected brains. Thepositive analogy is strong: similarity of form andinterrelationships, perhaps colour, a one-to-onecorrespondence of detail in the original to detailin the modelŠarbor vitae to arbor vitae, brachiapontis to brachia pontis. The negative analogy isunimportant, that is, it does not introduce mis-leading distortion for demonstration purposes.Plaster of Paris (or whatever material it may be) isnot the same as brain tissue; the brain model maynot be of the same size as an original brain, andso forth. Even three-dimensionality, however, isnot an essential part of the positive analogy, forsufficient representation can be obtained from atwo-dimensional brain-atlas with overlappingpictures of successive brain-levels (e.g. Krieg1957). This might be called a pictorial representa-tional model.That the symbolic, representational and verbalforms of model are all intimately related is wellknown in an advanced science like physics. "Ithas often happened in physics", say Einstein andInfeld, "that an essential advance was achievedby carrying out a consistent analogy between twoapparently unrelated phenomena" (1961: p. 270),The development of wave mechanics, begun byde Broglie and Schrodinger in 1925-26 is anexample of the achievement of a successful theoryby means of a deep and fortunate analogy. Simi-larity was proposed, and verified, in the unlikelyanalogy between the properties of a standing wavein an oscillating chord (e.g. a violin string) andthose of an atom emitting radiation. This "accous-tical analogy" (Einstein and Infeld 1961: p. 276)has become, with the special theory of relativity,the foundation of modern wave mechanics. In itsworking out, recourse must be had to mathematical(symbolic) models and sub-models, but its outlineis also simply illustrated with the pictorial andverbal models which occur in the work quoted.Einstein and Infeld carefully point out that anunderstanding not only of the similarities but ofthe differences of the analogy (i.e. the negativeanalogy) is important (p. 286),These illustrations bring out what seem to be anumber of important points. First, it is clear thatthe "constructed classes" of a model may besymbolically, verbally or even pictorially con-structed independently in the mind of the investi-gator, or they may be inductions from another,even unlikely, class of phenomena found in nature,with which the class of phenomena under investi-gation is to be compared by analogy. Second, animportant function of the positive analogy isevidently that its known dimensions in the modelare to be projected onto corresponding butunknown dimensions in the phenomena underinvestigation, so that new properties for experi-mental verification may be suggested about them.The known negative analogy is a stricture of theproperties in respect of which the two classes inanalogy cannot be compared. Again, the "model"is shown as a special form of theory, i.e. a syntheticgeneral statement (or connected series of state-ments) about what the relations between certainrecurrent factors can be expected to be (Reader1961: p. 211), but a theory stated by analogy.Finally, it appears that models of all kinds mayfulfil several functions: simplified representation(e.g. pictorial models), simplified interconnexion(e.g. symbolic models), simplified projection ofproperties (e.g. verbal models).METAPHORVerbal models, with which in anthropology weare mainly concerned, may be only metaphorical orfully analogic in character. In the earlier anthro-pological models, and more generally in theabsence of inductive information, the place ofanalogy tended to be taken by metaphor. Perhapsbecause the concept is so well known, the defini-tion of metaphor has not been easy. The OxfordEnglish Dictionary defines it as "the figure ofspeech in which a name or descriptive term istransferred to some object different from, butanalogous to, that to which it is properly applic-able". Aristotle ascribes the use of metaphor todelight in learning; Cicero traces it to the enjoy-ment of the author's ingenuity in overpassing theimmediate, or in the vivid presentation of theprincipal subject (Cope 1867). These views makemetaphor a decoration, an entertainment and adiversion as opposed to its logical use in the30dictionary definition. Ogden and Richards ap-proach it from both the logical and the emotivestandpoint: "Metaphor, in the most generalsense, is the use of one reference to a group ofthings between which a given relation holds, forthe purpose of facilitating the discrimination of ananalogous relation in another group" (1930: p. 213);and again, "(Indirect means of arousal are poss-ible) through the excitement of imagery (ofteneffected at low levels of refinement by the use ofmetaphor)Šused not, as in strict symbolizing, tobring out or stress a structural feature in a refer-ence, but rather to provide, often under cover of apretence of this elucidation, new sudden andstriking collocations of references for the sake ofcompound effects of contrast, conflict, harmony... or used more simply to modify and adjustemotional tone . . ." (p. 240).The philosopher Black, in a useful article(1954-55), has discussed three viewpoints onmetaphor. The first, which he calls the substitutionview, holds that a metaphorical expression is usedin place of some equivalent literal statement. If itis asked why this substitution should be made,one answer, apart from the entertainment oremotive view, is that metaphor is a kind of cata-chresis. This means the use of a word in somenew sense in order to remedy a gap in the vocabu-lary. If the metaphor thus serves a genuine need,the new sense it introduces should quickly becomepart of the literal language. Hence, on this viewnew metaphors should at least sometimes be inprocess of being absorbed (Black 1954-55: pp. 279-280).To hold that a metaphor consists in the presen-tation of an underlying analogy, or similarity, isto take what Black calls the comparison view.When Schopenhauer called a geometrical proof amousetrap, he was, according to this view, impli-citly saying that like a mousetrap, a geometricalproof offers a delusive reward, entices its victimby degrees, leads to a disagreeable surprise, andso on (Black: p. 283). It will be observed that thecomparison view is a special case of the substi-tution view: for it holds that the metaphoricalstatement might be replaced by an equivalentliteral comparison.The view which Black himself favours is theinteraction view. In Richards' words (1936: p. 93):"... when we use a metaphor we have two thoughtsof different things active together and supportedby a single word, or phrase, whose meaning is aresultant of their interaction". These "thoughts",according to Black, arise out of the system ofassociated commonplaces: the statements whichthe reader, or listener, would make as a layman,without taking special thought, of those thingswhich he held to be true about the class invokedby the metaphor. In this form, Black commits theinteraction view to the following seven claims. Forbrevity, they are set out in tabular form, with thepresent writer's criticisms in terms of the "com-parison" view alongside:"INTERACTION" VIEW OF METAPHOR1. A metaphorical statement has two distinctsubjectsŠa "principal" and a "subsidiary"one.2. These subjects are often best regarded as"systems of things" rather than "things".3. The metaphor works by applying to theprincipal subject a system of "associatedimplications" characteristic of the associatedsubject.4. These implications usually consist of com-monplaces about the subsidiary subject, butmay, in suitable cases, consist of deviantimplications established ad hoc by the user.CRITICISM Š"COMPARISON" VIEWŠWhat are these but members of two separate andallegedly parallel classes as compared in analogy?The "system of things" are surely the sets ofproperties defining the classes in analogy.The "associated implications" are surely proper-ties alleged to be held in commonŠi.e. the positiveanalogyŠbetween the classes being compared.In scientific metaphor they may have been care-fully thought outŠas a positive analogy.31"INTERACTION" VIEW OF METAPHORŠCont.5. The metaphor selects, emphasizes, sup-presses, and organizes features of theprincipal subject by implying statementsabout it which normally apply to the subsi-diary subject.6. This involves shifts in the meaning of wordsbelonging to the same family or system as themetaphorical expression; and some of theseshifts, though not all, may be metaphoricaltransfer. The subordinate metaphors are,however, to be read less emphatically.7. There is in general no simple "ground" for thenecessary shifts of meaningŠno blanketreason why some metaphors work and othersfail.CRITICISMŠ"COMPARISON" VIEWŠCont.i.e. selective positive analogy.This leads to an infinite regress of sub-metaphorswithin the meaning of the main metaphor, againstwhich Black barely defends himself.The sole ground is that the classes drawn inanalogy are appropriately or not appropriatelycompared: i.e. that their positive analogy is or isnot strong.Black believes that it is easy to overstate theconflicts between these three views, and that theymay all be true in different cases. But, he says,only the "reaction" view is of importance inphilosophy. The present contention is that only adeveloped form of the "comparison" viewpointŠwhat might be called the "analogic" viewŠisuseful in social science.Setting aside its emotive and decorative uses,metaphor would appear to be a summary form ofanalogy, but unexpanded: that is, with the positiveand negative analogies either uninvestigated, un-expressed, or not under control. Metaphor seemsan implied analogy in which one, or at most veryfew, striking items of positive analogy are seizedupon, without regard for the rest of the positive,and the probably substantial negative analogy.Conversely, logically unsatisfactory analogies,with disproportionately large negative analogies,are often uninvestigated metaphors expandedinto analogies by argument.Some of these points are well brought out by thecomments of Firth on Radcliffe-Brown's use of the"network" metaphor: "Now no one thought thathe meant that he was dealing with either a meshedfabric held together by knots, or an arrangementwith intersecting lines and interstices recallingsuch a fabric. We can take it for granted that likea modern painter, when he wrote network he wasexpressing what he felt by describing metaphoric-ally what he saw." (1954: p. 4). Firth is here bringingto light part of the substantial negative analogy ofthe class-term "network", which unfits it foranalogy with the arrangement of relationshipsbetween persons. He implies that the metaphorwas never meant to be expanded into an analogy,but is only of assistance in "seeing" a non-material situation in material terms. He goes on,"Bentham (1931) has a pertinent remark about thedanger of metaphors being used at first forillustration or ornament and afterwards made thebasis of an argument. .." (1954: p. 5).Since the negative analogy in metaphors, whenrevealed, is likely to be uncontrolled and high, themetaphor is not, without caution, suitable for usein models. It cannot be maintained, since analogyis involved, that the use of metaphor as or in amodel is logically invalid, but only that it is re-stricted and possibly misleading. Suitably chosen,it could make description and visualization ofreality more plain, provide some explanation of arange of interconnected facts, and even be asso-ciated with problem-orientation and prediction.It is suggested, however, that it would perform thelast two functions much better either whenexpanded into an analogy or when its implicitanalogy is known to be positive and dominant.Otherwise it functions as a model only in assistingthe operator to visualize that which is not visible,with some danger from the contra-indications ofits unknown negative analogy.In practice, many of the anthropological models32reviewed above consist of partially expandedmetaphors or incomplete analogies: these amountto the same thing. This is particularly true ofmodels based on analogy with the biologicalorganism. To begin with, the positive analogy ofthis model was artificially strengthened by select-ing for the "reality" side of the analogy onlysmall-scale, relatively homogeneous communities.These were obviously more closely analogous toan organism in equilibrium than large-scale urbangroups would have been. Even so, the positiveanalogy was limited and the negative analogy onlypartially expanded. In the positive analogy wereused the properties and relations of functionalinterdependence of parts and need-satisfaction,both resting on the presuppositions of closure andstable equilibrium. But the unexpanded negativeanalogy was formidable. The organic parts of abody were not like social persons in interrelation-ship. Their dispositions and functions (e.g. thefunctions of the kidney and liver) were entirelydifferent from those of a society. The surface ofthe body was not like the often arbitrary boundariesof a community, the metabolic cycle and physio-logical changes had no precise counterpart insociety, to speak of the sex of communities wasmeaningless, and so on indefinitely. Once the factof "functional" interrelationship in homogeneoussocieties was grasped and utilized, it is hardlysurprising that little further use was found for thisanalogy and the model it represented. It remains,if at all, as a metaphor, with the few items ofrelevant positive analogy and the unexpandednegative analogy characteristic of all metaphors.This analysis of the logic of metaphor hasperhaps gone too far in playing down the initialexploratory value of metaphors and their functionin stimulating new hypotheses. It is no doubt anatural sequence of thought to proceed from theknown to the unknown: to comprehend in onemetaphorical leap a similarity between what hasalready been conceived and that which has stillto be understood. The findings of this reviewsuggest, however, that to adhere literally to theimplied comparisons of a metaphor in the develop-mental phase, when the broad association hasalready been grasped, may well inhibit furtherinsight. If, through a large and unsuitable negativeanalogy, the metaphor does not bear expansioninto a model, it should surely either be discardedwhen its early heuristic purpose has been achieved,or its limitations should severely be borne in mindwhile no better model is available.In practice the matter is not as clear-cut as this.Metaphors in social anthropology seem to becapable of varying degrees of partial expansion,indeed there should be a continuum between thesuggestive but totally unexpandable metaphor andthe one which is so fruitful that it is virtually ananalogic model. Somewhere along this continuumcomes, for example, the "social network", ablyinitiated by Barnes (1954) and developed by Bott(1957) and later by Mayer (1961) and others. The"network" is "the total of ego's interpersonalrelations with other individuals." (Mayer 1961:p. 9). In Barnes' words: "Each person is ... intouch with a number of people, some of whom aredirectly in touch with each other and some ofwhom are not ... I find it convenient to talk of asocial field of this kind as a network. The image Ihave is of a net of points, some of which arejoined by lines. The points of the image are people,or sometimes groups, and the lines indicate whichpeople interact with each other." (Barnes 1954:p. 43). This metaphor has been useful in explainingclose-knit and loose-knit* community relations inNorway, England, and Black South Africa. It isdoubtful, though, whether the taxonomy of net-works recommended by Mayer (1962) would serveany useful purpose. That is taking the impliedanalogy too far.SUMMARYThe material of this review can finally be sum-marized as a set of considerations with respect tosocial models:1. The logical form of the model should beclearly understood: whether inductive (em-pirical) or deductive; whether ideal, idealizedor ideational; whether unithematic or poly-thematic; and whether verbal, symbolic orpictorial. Models should not be set up atmixed logical levels without due caution.2. Deductive idealized models seem appro-priate to the early stages of a discipline, toopen research in a little-known area. Theyalso apply in the developed stages whenmuch is already known inductively about thephenomena under review. For the inter-mediate stages of research, empiricalinductive models seem more fitting.ŁAnother metaphor. When metaphors are compounded in an implied analogy,its comprehensiveness for a given situation may only be that of the weakefmetaphor.333. Any model, however far "removed fromreality" should always permit of inductiveverification,4. A model should interrelate or "explain" thewidest possible range of data in terms of theminimal number of connected propositions,based on the simplest, fewest and mostfully acceptable presuppositions.5. It should be problem-oriented, and shouldpromote further hypotheses, or lead topredictions.6. The positive and negative analogies of themodel should be expanded and made mani-fest as fully as possible, so that the logicalvalidity of the model may be tested.7. A metaphor should not without caution beused in or as a model; and if it is used:7.1 it should be expanded into an analogy bymaking manifest its positive and nega-tive implied analogies; or if not soexpanded,7.2 its purpose in providing a tangibleconcept for intangible reality should beclearly stated, and its use restrictedaccordingly.8. Presuppositions of closure and stability insocial models are arbitrary. Some modelsshould also be based upon, and able tomeet, assumptions of:8.1 continuous social change unless dis-turbed;8.2 ubiquitous conflict, ambiguity and in-consistency;8.3 integration by constraint rather thanconsensus.Finally, this review should not be misconstruedas recommending the exclusive and pervasiveadoption of model-making as the only majortheoretical tool in social science. Kaplan (1964:pp. 277-287) has adequately listed for us theshortcomings of models: their over-emphasis onsymbols and on form, their oversimplification, theirfrequently misplaced rigour without deductivefertility, their misuse as "maps" of pictorialrealism, and the danger of the unconscioustransition from "that's what it is like" to "that'swhat it is". However, the dangers, as Kaplan says(p. 292), are not in working with models but inworking with too few which are too much alike; andabove all in belittling attempts to work withanything else.REFERENCES CITEDACHELIS, T. (1890) Die Geschlechtsgenossenschaft und die Entwicklung der Ehe. Zeitschrift der Gesellschaft furErdkunde zu Berlin, 25.BARNES, J, A. (1954) Class and Committees in a Norwegian Island Parish. Hum. Relat, VII, 1.BASTIAN, A. (1860) Der Mensch in der Geschichte. Leipzig.BEATTIE, J. H. M. (1959) Understanding and Explanation in Social Anthropology. Brit. J. Sociol., X, 1.BENTHAM, J. (1931) The Theory of Legislation, trs. R. Hildreth, C. K. Ogden, ed,, London, Int. Libr. Psychol,, Kegan Paul.BLACK, M. (1954-55) Metaphor. Proc. Aristot. Soc, New Series, 55.BOTT, ELIZABETH (1957) Family and Social Network: Roles, Norms and External Relationships in Ordinary UrbanFamilies. London, Tavistock Publications.BRAITHWAITE, R. B. (1953) Scientific Explanation: A Study of the Function of Theory, Probability and Law in Science.Cambridge Univ. Press.BUCKLE, H. T. (1857-61) History of Civilization in England. 2 vols., London.COPE, E. M. (1867) An Introduction to Aristotle's Rhetoric. Appendix B to Book III, Ch. II: On Metaphor. London.DAHRENDORF, R. (1958) Out of Utopia: Towards a Reorientation of Sociological Analysis. Amer. J. Sociol., LXIV, 2.DURKHEIM, E. (1919) Les Regies de la Methode Sociologique. 7*me 6dn, rev. et augm., Paris, Alcan.DURKHEIM, E. (1953) Sociology and Philosophy, trs. D. F. Pocock, intr. J. G. Peristiany. London, Cohen and West.EGGAN, F. (1950) Social Organization of the Western Pueblos. Chicago Univ. Press.EINSTEIN, A., and INFELD, L. (1961) The Evolution of Physics: The Growth of Ideas from Early Concepts to Relativity andQuanta. Cambridge Univ. Press.EPSTEIN, A. L. (1958) Politics in an Urban African Community. Manchester Univ. Press for Rhodes-Livingstone Institute.FIRTH, R. (1954) Social Organization and Social Change. Presidential address. J. Roy. Anthrop. Inst., 84, 1.FORTES, M. (1949) The Web of Kinship among the Tallensi. London, Oxford Univ. Press for International African Institute.FRAZER, J. G. (1890) The Golden Bough: A Study in Magic and Religion. London, Macmillan.FURNIVALL, J. S. (1948) Colonial Policy and Practice: A Comparative Study of Burma and Netherlands India. CambridgeUniv. Press.GLUCKMAN, M. (1942) Some Processes of Social Change illustrated from Zululand. African Studies, 1, 4.GLUCKMAN, M. (1948) An Analysis of the Sociological Theories of Bronislaw Malinowski. Rhodes-Livingstone PaperNo. 16.34GLUCKMAN, M. (1955a) Custom and Conflict in Africa. London, Oxford Univ. Press.GLUCKMAN, M. (1955b) The Judicial Process among the Barotse. Manchester Univ. Press for Rhodes-LivingstoneInstitute,HOGBIN, H. I. (1958) Social Change, Josiah Mason lectures delivered at the University of Birmingham. London, Watts.HOMANS, G. (1951) The Human Group, London, Routledge & Kegan Paul.KAPLAN, A, (1964) The Conduct of Inquiry. San Francisco, Chandler.KEYNES, J. M. (1921) A Treatise on Probability. London, Macmillan.KRIEG, W, J. S. (1957) Brain Mechanisms in Diachrome. 2nd edn., Brain Books, Evanston III.LEACH, E. R. (1954) Political Systems of Highland Burma: A Study of Kachin Social Structure. London, London School ofEconomics and Political Science.LEFEVRE, A. (1891) La Religion. Paris.LETOURNEAU, C. (1890) L'Evolution Juridique dans les Diverses Races Humaines. Paris.LEVI-STRAUSS, C. (1953) Social Structure in Anthropology Today: An Encyclopedic Inventory, A. Kroeber ed., ChicagoUniv, Press.LUBBOCK, J. (1855) Prehistoric Times, as Illustrated by Ancient Remains, and the Manners and Customs of ModernSavages. London.McLENNAN, J. F. (1897) Studies in Ancient History. New edn., London, Quaritch.MAINE, H, (1897) Ancient Law, its Connection with the Early History of Society and its Relation to Modern Ideas. 16th end.,London, Murray.MALINOWSKI, B. (1922) Argonauts of the Western Pacific. London, Routledge.MALINOWSKI, B. (1926) Crime and Custom in Savage Society. New York, Humanities Press.MALINOWSKI, B. (1935) Coral Gardens and their Magic. London, Allen & Unwin.MALINOWSKI, B. (1945) The Dynamics of Culture Change: An Enquiry into Race Relations in Africa. P. M. Kaberry, ed.,New Haven, Yale Univ. Press.MAUSS, M. (1896) La Religion et les Origines du Droit Penal. Retue de I'Histoire des Religions, 16.MAYER, P. (1961) Townsmen or Tribesmen: Conservatism and the Process of Urbanization in a South African City. CapeTown, Oxford Univ. Press.MAYER, P. (1962) Sociological Aspects of Labour Migration. Paper read at the Social Sciences Research Conference,University of Natal.MITCHELL, J. C. (1961) Tribalism and the Plural Society. Inaugural lecture given in the University College of Rhodesiaand Nyasaland. Oxford Univ. Press.MORGAN, L. H. (1878) Ancient Society, or Researches in the Lines of Human Progress from Savagery through Barbarismto Civilization. New York, Henry Holt.NADEL, S. F. (1957) The Theory of Social Structure. London, Cohen & West.OGDEN, C. K. and RICHARDS, I. A. (1930) The Meaning of Meaning. 3rd edn., revised. London, Kegan Paul.PARSONS, Y. (1951) The Social System. Glencoe, Free Press.PARSONS, T. (1954) Essays in Sociological Theory, Pure and Applied. Glencoe, Free Press.PERRY, W. J. (1927) The Children of the Sun: A Study in the Early History of Civilization. London, Methuen.RADCLIFFE-BROWN, A. R. (1922) The Andaman Islanders: A Study in Social Anthropology. Cambridge Univ. Press.RADCLIFFE-BROWN, A. R. (1952) Structure and Function in Primitive Society: Essays and Addresses, London, Cohen &West.READER, D. H. (1961) Anthropology as a Social Science. J. S.A. Nat. Inst. Personnel Res., 8.REDFIELD, R. (1955) Societies and Cultures as Natural Systems. J, Roy. Anthrop. Inst., LXXXV.RICHARDS, I. A. (1936) The Philosophy of Rhetoric. London, Oxford Univ. Press.RIVERS, W. H. R (1914) History of Melanesian Society. Vol. II, Cambridge Univ. Press.STEBBING, L. SUSAN (1930) A Modern Introduction to Logic. London, Methuen.STONE, J. R. and JACKSON, E. F. (1946) Economic Models with Special Reference to Mr. Kaldor's System. Econ, J., 56.TYLOR, E. B. (1865) Researches into the Early History of Mankind and the Development of Civilization. London, Murray.WAITZ, T. (1858) Anthropologie der Naturvolker. 6 vols., Leipzig.35