Durkheim’s Project for a Sociological Science

Mike Gane. Handbook of Social Theory. Editor: George Ritzer & Barry Smart. 2001. Sage Publication.

Durkheim did not claim to be the founder of sociology, but he did claim to have made a decisive modification to the intellectual tradition that originates from Saint-Simon and Comte. There is today considerable interest in the precise balance of influences on Durkheim’s thought and the complexity of his cultural and religious origins (see Strenski, 1997, which fills a gap in the standard biography by Steven Lukes, 1973). Discoveries of more texts from all periods of his career have opened up more lines of enquiry (e.g. Durkheim, 1996, 1997). These new materials are likely to be influential in making interpretation of Durkheim’s intellectual career more sensitive to the way Durkheim’s sociology intervened in the currents of French thought and politics in the period of the Third Republic, as they show the emergence of a sophisticated confrontation with the positivist tradition on the one hand, and idealist, neo-Kantian schools on the other (Gross, 1996). These reassessments are part of a considerable shift in the evaluation of Durkheimian sociology from one dominated by identifying Durkheim as a crude functionalist, either of the extreme left, centre or right, to one that recognizes the radical theoretical and political complexity of the Durkheimian project. It is this continuing interest in Durkheim which attests to the importance of the intellectual challenges and puzzles created in his sociological and philosophical writings.

Durkheim’s sociology belongs to a tradition which was evolutionist and holistic, and because of this refused to acknowledge a break between anthropology and sociology. It believed the role of social science was to provide guidance for specific kinds of social intervention. Crucial for the success of this project was for sociology to align itself with the other more established sciences within the institutional orbit of the higher educational system, the universities, a relatively autonomous haven for research which could remain free from direct political partisanship, particularly where scientific knowledge and practice arises within a democratic society basing itself on popular sovereignty. As part of this endeavour, the concept of social pathology was given considerable importance, revealing the extent to which Durkheim’s sociology was consciously modelled on the medical ideal of therapeutic intervention. There was no fundamental philosophical reason why sociology could not define and legitimate in its own special way social and moral norms. An essential part of Durkheim’s project was to work out specific rules for determining the distinction between normal and pathological social phenomena. In this respect Durkheim went much further in a direction outlined in principle by Comte and Spencer.

Although Durkheim secured a position teaching social sciences at the University of Bordeaux from 1887, his initial formation was in philosophy and he had taught philosophy at Lycées (Sens, Saint-Quentin and Troyes). His position at Bordeaux and later in Paris (from 1902) also required that he teach courses in education as well as sociology. The corpus of Durkheim’s work consists of major books (The Division of Labour in Society, The Rules of Sociological Method, Suicide and The Elementary Forms of the Religious Life), a small number of minor books (on Montesquieu, Primitive Classification (with Mauss), Moral Education), but also important lecture courses that were published posthumously (Socialism, Professional Ethics, The Evolution of Educational Thought in France) and courses which are based on notes taken by students (Pragmatism). There are also a large number of journal articles, some of which were gathered into collections, there are verbatim reports of debates, book reviews and letters (his letters to Marcel Mauss, for example, were published for the first time in 1998). But Durkheim did not work alone or in isolation, he founded a school of sociology organized around the journal calledL’Année Sociologique. The aim of the project thus organized was to ensure the implantation of sociology within the universities as a reputable discipline with a vital role to play in the identification of the normal forms of the emerging institutional structures of modern societies.

A notable feature of Durkheim’s sociological writing is its highly rationalist idioms. This is particularly clear in his methodological requirements, with their demand for clear and precise conceptual definition, rigorous formulation of problems, careful consideration for quality of evidence, canons of proof. The demand is that sociology should break with the prevailing ideological methods and establish a scientific rigour that can facilitate the discovery of the basic social laws governing social development and the formation of different modes of social solidarity. It is clear in this respect that Durkheim follows in general the aims outlined by Comte in his Cours de philosophie positive (1830-42), yet Durkheim criticized in detail the specific forms of Comte’s analyses and their results. It is very striking, for example, that Durkheim first of all focused on what he called ‘moral facts’ as against the emphasis on cognition in Comte’s law of the three states. Indeed, Durkheim’s studies reveal something missing in Comte: a serious absence of any sociological consideration of morals, laws, norms, sanctions, in the treatment of religion, culture and social development.

Durkheim’s emphasis seems to reflect a new awareness of the importance of Kant’s work and the problem of the relation between the individual and society, and can be seen as an attempt to unify within the sociological project the perspectives, transformed in a certain way, of both Kant and Comte (and Spencer). In Durkheim’s work on social categories, the sociology of knowledge and sociology of religion there is also a broader concern to deal with the philosophical problems raised by these two earlier writers and to place their ideas, suitably reconstructed, into a single analytical frame.

Durkheim repeated several times that he was ‘following the path opened up’ by Comte ([1907] 1980: 77), but was doing so in a specific way. He remained highly critical of Comtean epistemology and the way Comte’s law of the three states was formulated: ‘if, as Comte thought, historical development is unilinear, if it is constituted by a single and unique series which begins with humanity itself and continues without end, it is evident that, since all terms of comparison are lacking, it cannot be reduced to laws’ ([1906] 1980: 73). The law of the three social states, theological, metaphysical and positive, was not for Durkheim an adequate conceptualization of social species: it was essentially arbitrary—for why should evolution stop at the third state? Yet if Durkheim was hostile to the general form of Comte’s sociology he affirmed very specifically Comte’s analysis of European history since the middle ages: ‘Comte’s law correctly describes the way modern societies have developed from the tenth to the nineteenth century—but it does not apply to the entire course of human evolution’ ([1928] 1962: 268). And rather than adopt Comte’s terminology of the three states, he opted for a modified version of the morphological classification developed by Herbert Spencer, a classification by mode and degree of complexity of social composition, from segmentai to organized societal forms. Durkheim was at pains to insist the form was not unilinear, but involved genuine non-teleological diversity: ‘the genealogical tree of organized beings, instead of having the form of a geometric line, resembles more nearly a very bushy tree whose branches, issuing haphazardly all along the trunk, shoot out capriciously in all directions’ ([1888] 1978: 53). Thus Durkheim made a fine distinction between the genealogical analysis within a single real continuous society (historical sequence), a comparative analysis within a single society and societies of the same type, and a much more abstract comparative cross-cultural analysis leading to the construction of theoretical evolutionary typologies.

It is important to specify what is involved in these distinctions. Wallwork (1984) has attempted to specify a six-stage model of Durkheim’s evolutionary conception at the beginning of his career: the elementary horde, simple clan-based tribe, tribal confederation, ancient city-state, medieval society, modern industrial nation. In this perspective Durkheim divides these societies into two basic types, and in this respect he departs from Spencer: the societies that are ‘mechanical’ accretions of elements (kinship groups) as ‘compound’ or ‘doubly compound’ in Spencer’s terms, as against those societies that are organized on the basis of an ‘organic’ division of labour and interdependent specialization of function (comprising medieval society and the modern nation). The idea of basing a classification on what is called social morphology, or what has become since Spencer known as social structure, is quite different from one based on cultural configuration such as can be found in Comte’s sociology. For Comte, rather as for Marx, it was essentially one decisive element in society, its determining element, the method of knowing, which should form the basis of classification. Comte’s classification, in effect, followed a parallel five-stage model: fetishist, polytheist, monotheist, metaphysical and positive societal types, in a single logical sequence explained by the struggle of positive reason against theological reason. Marx, of course, also had a five-stage model: primitive communism, slave, feudal, capitalist and higher communist societies, a sequence determined by the economic mode of production and class struggle.

There are notable differences between the models of Marx and Comte on the one hand, and Spencer and Durkheim on the other. In the first place, for Marx and Comte the final stage is a messianic logical construction of a future Utopian state. For Spencer and Durkheim the sequence contains no reference to the future, which remains in principle open and unknowable. The Spencer-Durkheim scheme is descriptively and empirically holistic: it does not depend on a presumed partwhole form of causal theory. Thirdly, when theory is introduced, both Spencer and Durkheim reject the idea that there is a simple development of forms of social inequality through the sequence. Such stratification is affected not only by degree of social complexity, but by the distribution of power in society and this is decisively determined by the condition and form of social mobilization: for war, external or civil, or peace and industry. In other words, Spencer and Durkheim reorganized the Comtean evolutionary thesis that society moves from warlike to industrial occupations, to one in which these can be found as modal states at any stage of development.

Problems of Method

Following in Comte’s footsteps, Durkheim attempted to specify the unique domain of sociological phenomena within the branches of the scientific division of labour. It is tempting to suggest that this is an attempt to align sociology within the rationalist tradition stemming from Bacon and Descartes. But Durkheim insists that rules of method are strictly parallel with other ethical and moral rules: ‘methodological rules are for science what rules of law and custom are for conduct,’ and whereas the natural sciences appear to form common ground, the ‘moral and social sciences’ are in a state of anomie: ‘the jurist, the psychologist, the anthropologist, the economist, the statistician, the linguist, the historian, proceed with their investigations as if the different orders of fact they study constituted so many different worlds’ ([1893] 1964: 368). Rules of method are far more than rules for gathering information. They organize and regulate the field and identify the ground for scientific strategies and legitimate the way these strategies should be assessed. Durkheim’s writings on methodology, not surprisingly, have given rise to important controversies. These are of two kinds. The first concern the orientation and content of some of the rules. The second the apparent conflict between certain rules.

The methodological orientation is from the start one which strives for maximal objectivity of investigation. As if traumatized by a surfeit of individual and subjective introspection, even the primary definition of the terrain of sociology itself, the very object of social analysis, Durkheim formulated as ‘any way of acting, whether fixed or not, capable of exerting over the individual an external constraint’ ([1895] 1982: 59). In this argument, The Rules of Sociological Method seems intent on correcting an apparent weakness in previous sociological practice, for the emphasis is insistent: all preconceptions must be discarded, never assume the voluntary character of a social institution, always study social phenomena which are detached from individual forms. The fundamental rule for observing social phenomena is ‘to consider social facts as things’ ([1895] 1982: 60). It is evident that Durkheim considered the common ideological formation of the sociologist as modern citizen in a sense, and paradoxically, both a support and a fundamental obstacle to social research. His critical judgement is that ‘instead of observing, describing and comparing things, we … reflect on our ideas … Instead of a science which deals with realities, we carry out no more than an ideological analysis’ (p. 60). It is no wonder that this particular text has been read as a revolutionary manifesto in sociology, for it essentially demands that each sociologist reverse habitual and everyday forms of thought. In this sense it is a fundamental text in a tradition that includes Comte and Marx as central figures: social science comes into being in a revolutionary, epistemological break with ideological methods.

The very specific character of Durkheim’s method is that it carries the idea of objectivity into the very conceptualization of its object, the social itself. The terms social exteriority, constraint, collective, are transcendent as against individual manifestation, and are clearly conceived as decisive for the investigator and for social research in general: the most objective of objects are the fixed social forms and these are to be given priority because they offer a privileged route for investigative analysis. Durkheim is, however, very clear and explicit: these terms are provisional. They are to be adopted in the first instance because they offer the best chance of avoiding significant obstacles to social science. It is clear that as a manifesto, this system of rules does call for a revolutionary transformation in the mode of practice of sociologists, and calls for a commitment to sociology as a vocation. This vision is not concerned with the isolated and lonely genius, but with a collective practice, organized and disciplined within the modern educational system. Sociological research is a collective procedure, its ideas are subjected to rational rules of evidence and verification, unique to sociology. These rules mark out and define a territory for study against competing claims: the domain of social facts.

It is clear, however, that these rules are also closely bound up with Durkheim’s own ideas about the content of social analysis, and indeed his own changing and developing research priorities: what seems at first to be an attempt to produce a definitive text on methodology, The Rules of Sociological Method, is more complex than its appears. Many of its major formulations had already appeared as the first Introduction to The Division of Labour in Society (edition of 1893). This Introduction was replaced in the second edition of 1902, but can be found, as an appendix, in the English translation of 1964 (but not in the second translation of 1984). The first formulation of the field of study is developed in terms not of social but moral facts: ‘moral facts consist in a rule of sanctioned conduct’ (1964: 425). Durkheim adopts this definition since the study of moral evolution no longer depends on the study of norms but also of sanctions which he conceives as ‘an external fact reflecting [an] internal state’ (p. 425). His aim remains consistent: the development of the ‘positive science of morality is a branch of sociology, for every sanction is principally a social thing’ (p. 428). In this first Introduction there is a very clearly demarcated separation between those forms of action which are moral obligations and other ‘gratuitous acts’ which are at the free choice of the individual and which Durkheim assigns to the domain of aesthetics and art (pp. 430-1).

What happens between this first Introduction to The Division of Labour and the appearance of the essays called The Rules of Sociological Method which were published separately as articles in 1894 (before appearing in modified book form in 1895), is that all the characteristics of the moral fact are displaced and incorporated into the category of the social fact. It is significant to note for instance that Durkheim in 1893 referred to a group of acts as ‘gratuitous’ and outside morals: ‘the refinements of worldly urbanity, the ingenuities of politeness … the gifts, affectionate words or caresses between friends or relatives, up to the heroic sacrifices that no duty demands … The father of a family risks his life for a stranger; who would dare to say that was useful’ ([1893] 1964: 430-1). In The Rules, however, Durkheim specifies a whole new range of ways in which external constraint occurs from informal sanctions of ridicule and social ostracism to technical and organizational necessities. Moral obligation and sanction form the first object of Durkheim’s methodological reflection, but constraint and circumstantial necessity broaden this into the essential characteristics of social facts in general. In these 1894 formulations it is made clear none the less that the sociologist must choose to give primacy to the study of facts from ‘a viewpoint where they present a sufficient degree of consolidation’ (Durkheim, 1988: 138). By the time the book version of The Rules was published in 1895 with a large number of revisions, this particular injunction was changed to the rule that the sociologist ‘must strive to consider [social facts] from a viewpoint where they present themselves in isolation from their individual manifestations’ (1982: 82-3). It seems clear that Durkheim had altered his research priorities and was edging towards the full scale study of suicide statistics, published in 1897 not long after The Rules. Towards the end of The Rules, Durkheim gives suicide as an example of a problematic social fact: ‘if suicide depends on more than one cause it is because in reality there are several kinds of suicide’ (p. 150). The existence of suicide statistics makes it possible for sociologists to study social currents as phenomena that are ‘independent of their individual manifestations.’ Because the study concerns social rates of suicide there is something more to the phenomena than random ‘gratuitous acts.’ It is apparent then that far from being a fixed and unique definition of the object of sociology, Durkheim in fact has a number of options which are refocused and specified according to the task at hand. He does not seem to be worried if he does not follow to the letter his own hastily conceived prescriptive rules, as long as he remains consistent at a higher level of epistemological theory.

There was a strong injunction throughout the methodological writings to this point (1895), that the sociologist must start from things not ideas, indeed the sociologist must start from ‘a group of phenomena defined beforehand by certain common external characteristics’ ([1895] 1982: 75). This strategy demands analysis works from these external features towards an understanding of their internal causal relationships. When investigating suicide statistics Durkheim seems to have started in this way as prescribed by his rules. The procedure required that he group together those suicides with the same external features (for example, how suicides were committed) and ‘would admit as many suicide currents as there were distinct types, then seek to determine their causes’ ([1897] 1970: 146). In the analysis published as Suicide ([1897] 1970), Durkheim says this procedure could not be used, rather he was ‘able to determine the social types of suicide by classifying them not directly by their preliminarily described characteristics, but by the causes which produce them’ (p. 147). Durkheim uses what he calls ‘this reverse method.’ Instead of proceeding from the characteristics in the facts themselves, he says ‘once the nature of the causes is known we shall try to deduce that nature of the effects … Thus we shall descend from causes to effects and our aetiological classification will be completed by a morphological one’ (p. 147). It comes as something of a surprise to learn that the social causes of suicides (anomic, egoistic, altruistic) are already known, since the concept of altruism (of Comtean origin of course) has hardly figured in Durkheim’s sociology up to this point. It is also surprising that Durkheim could reverse the order of analysis, prescribed so insistently in his recent writings, with such ease and assurance. In the study of suicide rates Durkheim does not group suicides according to an external characteristic of the act of suicide as his previous rules required: he classifies them according to his theory of the major social causes of suicide.

Apart from the problems of orientation and inconsistency of usage, there is also a profound problem concerning the relation of theory to method: are the two independent or dependent? Clearly Durkheim did not think method completely separate from theory but developed alongside the progress of substantive sociology itself. At least in one crucial instance a conflict between a substantive thesis and methodological principle can be identified. This conflict arises in the central chapter of The Rules, in a discussion which deals with the problem of determining the difference between normal and pathological social facts. It is clear from the social analysis presented in The Division, and especially its famous second Introduction to the edition of 1902, that Durkheim considered French society to be in a grave condition of malaise, due to a social structural abnormality: with the abolition of the guild system in the eighteenth century a severe structural imbalance had been introduced into French society. The severe political oscillations which had occurred in French society since then were ultimately a result of this social structural imbalance: only by reintroducing some modern equivalent of the guilds, towards a pattern he called an ‘institutional socialism,’ could the normal system of counter-weights be restored.

The methodological problem posed at the heart of the Durkheimian project was to define unambiguously the way such a question could be resolved. His primary rule is clear: if the social fact is general in the average form of the social species under consideration the fact is to be judged normal. But he introduced another much more theoretical consideration, requiring a demonstration for a normal fact ‘that the general character of the phenomenon is related to the general conditions of collective life in the social type under consideration’ and this ‘verification is necessary when this fact relates to a social species which has not yet gone through its complete evolution’ ([1895] 1982: 97). These two rules tend to contradict each other in the case under discussion. If the anomic forms of the division of labour are a continuation of a feature of segmentai society, and ‘now increasingly dying out, we shall be forced to conclude that this now constitutes a morbid state, however universal it may be’ (p. 95). In other words the theoretical analysis of the forms and functions of regulation suggest that economic anomie is pathological even if it is general; the principal rule would suggest its generality indicates normality. This problem goes to the heart of modern Durkheimian scholarship. On the one hand, those Durkheimians who follow the primary empirical rule would be forced to conclude that the modern democratic state with its absence of guild forms has proven to be a normal social type. On the other, those Durkheimians who follow more strictly his theoretical analysis of modern societies with their lack of solidarity see a proliferation of many kinds of social pathological phenomena.


Durkheim’s basic theory developed and changed over the course of his career, as did his methodological reflections. In the earlier writings he held that primitive societies were characterized by similitudes and passions, while the advanced societies by individualism and calm restraint. He later revised this view completely. One example of this can be seen in his writing on education. In his early lectures on moral education (1973) he argued that there was a plague of violent punishment in the schools of the middle ages, and the lash remained in constant use up until the eighteenth century. After researching his lectures on educational thought in France ([1938] 1977), he describes the idea of the violent medieval colleges as simply ‘a legend.’ The reason for this, he argues, was that the educational communities remained essentially democratic and these forms ‘never have very harsh disciplinary regimes’ because ‘he who is today judged may tomorrow become the judge’ ([1938] 1977: 155-7). The new analysis suggests that the turn towards a more oppressive disciplinary regime began at the end of the sixteenth century, just at the moment when the schools and colleges in France became centralized and cut off from the outside community. In these circumstances, he says, the whip became a regular feature of college life.

In his lectures to teachers Durkheim discussed the problem of how to arrive at a rational approach to discipline and punishment in school. Between the offence and the punishment he observed there is a hidden continuity, for they are not ‘two heterogeneous things coupled artificially’ ([1925] 1973: 179). Because the mediating term is obscure, a series of misleading theories of punishment arises. One such misleading theory sees punishment as expiation or atonement, another sees it primarily as a way of intimidating or inhibiting further offences. From a pedagogical point of view the problem concerns the capacity to neutralize the demoralizing effects of an infringement of group norms. The true objective of punishment, he argues, is a moral one. Its effectiveness should be judged by how far it contributes to the solidarity of the group as a whole. The problem is that certain kinds of punishment can contribute to the creation of further immoral acts (p. 199), and once applied, punishment seems to lose something of its power. A reign of terror is, in the end, a very weak system of sanctions, even driven to extremes by its own ineffectiveness. The recourse to corporal punishment seems to involve a counter-productive attack on the dignity of the individual, a dignity valued and fostered in modern societies.

The central theoretical issue here was addressed once again in his attempt to reconstruct the theses of The Division of Labour in Society, in an article of 1900 called ‘Two Laws of Penal Evolution’ (1978: 153-80). Durkheim criticizes Spencer for thinking that the degree of absoluteness of governmental power is related to the number of functions it undertakes, but he works towards a very Spencerian formulation: ‘the more or less absolute character of the government is not an inherent characteristic of any given social type’ (p. 157). Indeed, it is here that Durkheim presents an account of French society that can be seen to be diametrically opposed to that of Marx: ‘seventeenth-century France and nineteenth-century France belong to the same type’ (p. 157). To think there has been a change of type is to mistake a conjunctural feature of the society with its fundamental structure, for governmental absolutism arises, not from the constituent features of a social form, but from ‘individual, transitory and contingent conditions’ in social evolution (p. 157). It is this very complication which makes analysis of social type and analysis of the logic of this type taken with the specific form of governmental power extremely complex, since transitory forms of power can neutralize long run social organization.

Durkheim’s argument suggests social theory is often mistaken in thinking the state as either a purely repressive machine, or that the purely political division of powers can deliver political and social liberty in the fullest sense. For Durkheim, the thesis that freedom is freedom from the state ignores the fact that it is the state ‘that has rescued the child from family tyranny [and] the citizen from feudal groups and later from communal groups.’ Indeed, Durkheim argues the state must not limit itself to the administration of ‘prohibitive justice … [it] must deploy energies equal to those for which it has to provide a counter-balance’ ([1950] 1992: 64). Against the political illusion of power, for example as found in Montesquieu, Durkheim in effect tries to show that liberty is based on a particular form of the total social division of power: the state ‘must even permeate all those secondary groups of family, trade and professional association, Church, regional areas, and so on …’ ([1950] 1992: 64) if the full potentialities of human development are to be realized.

But the state can become too strong and develop its own pathological dimensions and capacities. In his pamphlet ‘Germany Above all’: German Mentality and War (1915) he presented a critique of the ideas of the German political philosopher Heinrich Treitschke, which he took to be representative of the mentality that brought war not just to France and Europe in 1914. He was careful in fact to say that he was not analysing the causes of war, only one of the manifestations of a condition of social pathology (1915: 46). Durkheim contrasts the democratic idea in which there is a continuity between government and people with Treitschke’s thesis that there is a radical antagonism between state and civil society. This latter idea requires a state power capable of enforcing a mechanical obedience from its citizens—their first duty is to obey its dictates—and leaders who are possessed of enormous ambition, unwavering determination, with personalities characterized by aspects that have ‘something harsh, caustic, and more or less detestable’ about them (pp. 30-4). In practice these states flout international law and conventions, and their idea of war pushes the development of military technologies which are almost ‘exempt from the laws of gravity … [t]hey seem to transport us into an unreal world, where nothing can any longer resist the will of man’ (p. 46).

This analysis of the German war mentality draws on Durkheim’s crucial concept of anomic states, developed most clearly with respect to anomic forms of the division of labour ([1893] 1964: 353-73) and anomic suicides ([1897] 1970: 241-76). It is evident from Durkheim’s first formulations of this idea that it is derived from Comte, who developed it in relation to an analysis of the unregulated division of the modern sciences. Durkheim also follows Comte’s conception (itself derived from Broussais) of pathological facts as exceptional phenomena, that is exhibiting exceptionally high or low intensities. Thus crime, for example, is not in itself an abnormal feature of human societies: the sociologist has to determine normal and abnormal rates of crime. Changing intensities of social facts in Durkheimian theory are determined in relation to modifications in the power dynamics of social systems. Where there is a shift towards the concentration of power in the state, as occurs in wartime, the structures protecting individual values are weakened. In wartime there is to be expected not only an increase in altruistic suicide, most commonly associated with military organization (1970: 228ff), but also an increase in civil homicides since the individual as such is less protected in moral value (1992: 110-20).

It seems clear that there is a long-term continuity in Durkheim’s interest in moral statistics from the early essay on variations in birth rates through a range of studies of family, divorce, to political statistics (Turner, 1993). This aspect of his sociology has attracted attention as installing an experimental rationalism as a founding moment in the modern discipline (see Berthelot, 1995: 75-105). But this is to overlook the dependence of this methodology on the theoretical frame derived in large measure from Comte, and acknowledged by Durkheim. Whereas Comte focused on a single line of evolution unifying the historical experience of humanity as a whole, Durkheim investigated the dispersive branchings of social evolution and this strategy may have legitimized a more experimental and comparative methodological inventiveness. There is also a marked difference in the conception of the role of theory in sociology between Durkheim and Comte. For the latter, the aim of analysis is to be able to construct a hierarchical system of laws of co-variation: no reference to causal explanation is required; for Durkheim the role of theory is paramount in the search for causes, and is essential for a complete sociological explanation of social laws. Durkheim, it must be stressed, still embraced the aim of discovering basic social laws, and many of these are formulated in his early works. These are always related to a causal or aetiological analysis, which even becomes the explicit organizing principle, as is the case in Suicide.

Durkheim’s theoretical frame did not remain static, as has already be indicated. The most evident development of theory in Durkheim’s work can be found by comparing the depiction of early societies in the two studies at each end of his career: The Division of Labour in Society ([1893]) and The Elementary Forms of the Religious Life ([1912] 1995). In the first study the fundamental fact of the early societies is that they are held together by ‘bonds of similitude’ and characterized by intense and violent reactions to infringements of the highly uniform ‘collective consciousness.’ There is little in the way of social differentiation, even the gender division of labour is so slight there is no contractual regulation between the sexes. The era of ‘mechanical solidarity’ was one of sexual promiscuity ([1893] 1964: 57-8). Durkheim’s investigations into Australian tribal society led him to change this view fundamentally. He came to see kinship organization as complex, and based on deeply embedded forms of sexual and age divisions. He came to see social structure as the complex outcome of symbolic practices, particularly those crystallized in ritual traditions. He established the thesis that ritual beliefs were structured on knowledge categories which were socially produced and reproduced. Fundamental to such systems of religious categories were the concepts of the sacred and profane, good and evil, which were involved both in organizing such rituals and being at the same time produced by them. Instead of elaborating or criticizing Comte’s theory of early societies as being characterized by forms of fetishism (worship organized in relation to charged objects), Durkheim suggested that the earliest form of religion was totemistic (group kinship and religious practices were organized in relation to a hierarchy of objects: the totemic emblem, the totemic group, the totemic species). The practices of the group produced widely different forms of experience, for example, religious effervescent, high-energy ceremonials contrasted with low-energy utilitarian food gathering. These socially produced distinctions formed the material basis for category differentiation. In this way Durkheim thought he could arrive at a definitive sociological critique of Kantian a priorism on the one hand, and Spencerian individualism on the other. Durkheim tried to identify those groups which could draw moral strength from the solidarities produced by sacred rituals, and those with weaker solidarity who would then be vulnerable to the process of scapegoating, for example misfortunes befalling the group were blamed on women ([1912] 1995: 404).

Durkheim developed a theory of the fundamental importance of gender as he encountered the materials on Australian totemism. These also led to a theory of sacred categories, of good and evil, on top of the distinction between the sacred and profane social spheres, a distinction which Durkheim showed to be drawn in ritual practice. This investigation also tried to show that the idea of the individual soul was intimately linked to the structure of social groups and their internal differentiation (in some groups for example, women did not have souls). Because Durkheim’s attention had shifted to these symbolic processes and practices of intervention in and reproduction of such symbolic materials, it has been assumed in some interpretations that his whole sociology had itself become a subjective exercise in symbolic interactionism (Stone and Farberman, 1967). It seems clear from the text of The Elementary Forms, in fact, that there was no break in continuity of methodological reflection and prescription. However, the focus of analysis was no longer on the transcendent external modes of sanctioned conduct (moral facts), but had moved to social epistemology, or what he called ‘the sociology of knowledge and religion’ which examined the way immanent infrastructures imposed their exigencies on action. In other words, Durkheim had moved to a large scale and empirically based study of the cognitive structures of the earliest societies: Comtean terrain, but Durkheim locates his discussion almost entirely with respect to post-Comtean theory, and in a very different theoretical strategy.

The Current Situation

If we review the reception of Durkheim’s work among anthropologists and sociologists it is very apparent that there has been a good deal of theoretical confusion as to the precise nature and meaning of Durkheim’s methodology. The adoption of structural functionalism in America showed that the methodology could be adapted for social analysis in an advanced democracy. But there was a reaction which can be seen in Lukes’ extreme introduction to the second English translation (1982). More recent discussion, particularly the balanced accounts in the conference proceedings of the centennial meeting in Bordeaux (Cuin, 1997), point to the epistemological sophistication of Durkheim’s interventions.

Method is increasingly seen as the key to understanding the strategic connections in Durkheim’s sequence of studies. Against this continuity, the studies themselves have come under considerable substantive criticism. The Division of Labour in Society, for example, came under considerable critical attack in the writings of Sheleff (1975), and Lukes and Scull (in Durkheim, 1983). Suicide came under severe criticism by subjectivist trends in sociology, particularly social constructionists, from Douglas (1967), Pope (1976) to Atkinson (1978). The Elementary Forms of Religious Life has had an equally problematic career. After the text appeared, its central theses were frontally attacked by a range of writers. Adam Kuper has reviewed these criticisms and concludes that the ‘model of a segmentary structure based on unilineal descent groups is a sociological fantasy’ (Kuper, 1985: 235). Other critics have argued that the whole problematic of totemic society is as false as that of Comte’s theory of fetishism. The paradox is, therefore, why does Durkheim’s work remain significant? Kuper suggests the ‘apparently paradoxical fruitfulness of Durkheim’s work, despite its substantive failure, is … due to the power of certain elements of his methodology; to the importance of some of the questions which he set on the agenda of the next generation; and, above all, to the sense which he communicated of the richness, complexity and sociological interest of ethnographic materials’ (1985: 235).

From the 1960s different styles of Durkheimian scholarship were pursued against the trend of Durkheim criticism. These included most notably the work of Jeffrey Alexander in cultural theory (1988), Frank Pearce in politics and law (1989), Steve Taylor on suicide (1982), Mike Gane on method (1988) and institutional socialism (1992), Jennifer Lehmann in gender theory (1994), W.S.F. Pickering in the sociology of religion (1984), Mark Cladis on ethics and liberalism (1992), W. Watts Miller on ethics and politics (1996), S. Mestrovic on postmodern culture (1988, 1991, 1992) and Warren Schmaus on methodology (1994). There were a number of thematic collections which attempted to address certain themes; collections edited by Stephen Turner on religion and morality (1993), Allen, Pickering and Watts Miller’s collection on The Elementary Forms (1998). Further collections on Suicide, and on Collective Representations, are in Press. Pickering and Martins’ Debating Durkheim (1994) addressed a wide range of issues in dispute. Many of these debates and publications have been organized by the British Centre for Durkheimian Studies, directed by W.S.F. Pickering, at Oxford. During this period there have been a number of introductions to Durkheim’s works which have been far more positive than that of Anthony Giddens, particularly those by Ken Thompson (1982), Robert Alun Jones (1986) and Frank Parkin (1992). A move towards a rapprochement between the Weberian and Durkheimian traditions is now well under way (see Boudon, 1995).

It seems clear in retrospect, however, that Durkheim’s work entered in the 1960s a period of intense critical scrutiny after being, in the United States particularly, the dominant sociological paradigm (see Parsons, 1937 and Merton, 1949). The point at which the rejection of Durkheim’s method and theory reached its height was perhaps at around 1980 with three essays: a ‘Critical Commentary’ on Durkheim’s work by Anthony Giddens (Giddens, 1978: 101-20), an assessment of the adequacy of Durkheim’s theory as an ‘integrated sociological paradigm’ (Ritzer and Bell, 1981), and the vitriolic introduction by Steven Lukes to the second (1982) translation of The Rules of Sociological Method. In the 1990s a series of conferences and theme-based publications produced a complex fin de siècle balance sheet. At first sight the question might seem to be whether these publications mark the end of a period, or announce a rebirth of Durkheimian perspectives. But the issue is probably more complex and less heroic. On the one hand, after a period in which the core sociological problems were related to Marx, Durkheim and Weber (see Morrison, 1995), there is likely to be a recovery of the wider band of influences on social theory, back into the eighteenth century as a way of contextualizing the irruption of sociology at the beginning of the nineteenth century (Comte, Mill, Littre, Spencer), and a wider appreciation of the theorists at the beginning of the twentieth century (including Pareto, Tönnies, Simmel and others). Twentieth-century sociology is marked by a division arising out of the legacy of the Durkheim school: the direction taken by structural modernist sociology, and another direction, influenced by Georges Bataille, leading towards postmodernism.

What is the significance of the Durkheimian legacy in this picture? Although Durkheim himself stressed the role of decisive discoveries in the progression of a science, it seems there is another important critical function in the growth of sociological knowledge. As against making discoveries this might be called learning from errors. It might thus be the case, and paradoxically, that Durkheim’s empirical analyses have been shown to be at best only partially successful. The great strength of these studies was that they were experimental, and in important respects Durkheim often tried to improve his analyses where they became evidently untenable. He did this with great theoretical skill and inventiveness, always aware of the importance of reconstructing his conceptual schemes and methodological controls. His analytic scheme is not confused with empirical evidence and this makes it possible to check and correct, or reject his analyses. This process has become an essential element of modern sociology. It was not inaugurated by Durkheim, whose treatment of Comte for example was not exemplary. It was inaugurated by scholars working on the Durkheim corpus, and became an essential moment in the formation of all modern sociologists.