1Émile Durkheim showed an early interest in religious phenomena. Some commentators, therefore, have seen a continuity and unitary trajectory in his sociology of religion.  Others, however, have stressed the discontinuity, taking seriously what Durkheim said in 1907 when he contradicted Simon Deploige – namely, that 1895 had been the turning point for him, the year of his famous “revelation” (1907a).  Both views rely on good arguments, probably because they both reflect part of the reality. However, one must define the words “continuity” and “discontinuity”. From the very beginning, Durkheim centered his research program on four questions:
- the question of definition (“What then is religion?” [1887b, 301]);
- the sociological question proper (the manner in which “religion as a sociological phenomenon should be studied” [1887a, 149]);
- the question of the relationship between religion and morality (“the similarities and differences which exist between moral and religious commandments” [1887a, 163]); and
- the question of the future of religion (“how religions are in the process of disappearing and what will be left of them” [1887a, 164]).
2It is from this angle that we will reread Durkheim’s work from 1886 to 1906-1907, the year of his course on the origins of religious life (Durkheim 1907b), in which the structure of The Elementary Forms of the Religious Life was already being outlined in various aspects. In doing so, we will follow a chronological order, which seems to us to be best suited to avoid the risk of analyzing Durkheim’s research into religion solely as a series of preparatory stages toward his final book. We will try instead to reread Durkheim’s work prior to 1912 as though we did not know he was the author of The Elementary Forms. We might ask ourselves to what extent this approach would allow us to better understand Durkheim’s book, or to understand it in a way other than the one we are used to.
1 – The Early Texts Until The Division of Labor in Society
3Durkheim first dealt with the theory of religion when examining the works of Herbert Spencer (1886a), Wilhelm Wundt (1887b) and Jean-Marie Guyau (1887a). These three analyses made up a coherent whole which was first and foremost characterized by the centrality of the question of definition (a). According to the three authors considered by Durkheim, the elements that defined religion were: the idea of a supernatural being and of mystery (Spencer 1886a, 185, 187); the feeling of an ideal existence that was expressed by the gods (Wundt 1887b, 302); the gods again, this time seen in their relationships with humans, which religion had the task of regulating (Guyau 1887a,?150). In Durkheim’s view, these definitions, with the partial exception of the one put forward by Guyau, contained the same error: they suggested that religious beliefs were the product of an individual mental operation, “a simple event of the individual consciousness” (1887a, 149). Thus, Durkheim saw numerous versions of a definition model being outlined. According to this model, religion was defined starting from a basic psychological fact: fear or wonder, a need to understand, or a yearning for the ideal. Seen from this perspective, any introduction of explanatory social factors into the study of religion would be inappropriate or superfluous.
4According to Durkheim, the error made by psychological types of definition was to stop at the object of religious beliefs (supernatural beings, deities), an object which could rightly be seen to be the result of individual mental elaboration (whether it be imagination, projection or reification). In order to avoid this error, Durkheim chose to leave the object of religious beliefs aside and concentrate on their modality. “Religious” is not the thing believed, but the way in which one believes. For Durkheim, the modality of belief that defines religion is “faith”, that is, “all belief which one accepts or is subjected to without discussion” (1886a, 195). This was a formal definition which raised a few questions. Durkheim left aside the object of religious beliefs in order to highlight the social aspect of religion, but it is not immediately clear why faith should involve a social dimension rather than just consist of individual actions or states of consciousness. In fact, it was Durkheim himself who begged the question and stated that the faith he was interested in was not faith in general but a “common faith”, that is, of a social kind. Furthermore, even if we were to admit that every religion is a faith, it does not necessarily follow that every faith is a religion. Taken on its own, the formal criterion of the modality of belief is insufficient and vague. Nevertheless, Durkheim took his time to give up on this idea.
5By way of making up for the weakness of the first Durkheimian definition of religion, we might point out that the obligation which is at work in faith (the latter is something which “one accepts or is subjected to without discussion”, our emphasis) needs the support of a coercive force which only society can exert. Indeed, the idea of obligation was at the heart of Durkheim’s initial thoughts on religion. Nevertheless, obligation is a legal or moral notion whose application in the area of religious phenomena does not follow automatically. Consequently, it was the question of the relationship between religion and morality (c) which drew Durkheim’s attention in his writings during the 1880s. According to him, religion and morality implied obligation to the extent that they both expressed social sentiments: the problem, then, was to understand what distinguished religion from morality once we left aside the objects of religious beliefs (gods, spirits, etc.), which were nonessential. It was particularly in Wundt’s Ethik (published in 1886) that Durkheim found an idea about the way in which the transition from religion to morality could take place: indirectly, by means of customs. However, Wundt saw this transition as an evolution from inferior to superior: religion was merely the symbolic and rudimentary manifestation of moral notions, which were destined sooner or later to express themselves in their pure, i.e. nonreligious, form. This was another solution which continued to seduce Durkheim until his Moral Education (1898-1899). However, it is clear that one of the strong ambitions of his sociology of religion was to attempt to explain religious phenomena on the basis of their specificity: just as it was wrong to say that they consisted of inferior forms of knowledge, similarly, we could not explain them by reducing them to perceptible symbols of moral law. Faced with this problem, Durkheim contented himself – for the time being – with distinguishing between religion and morality by distinguishing between ideas and practices: religious obligation was fundamentally concerned with ideas, or beliefs, while moral obligation was concerned with practices. This broke with the Kantian moral philosophy of the time, whereby the normativity of moral obligation was to be found at the inner level of conviction rather than the external level of practices. Furthermore, from this perspective, the ritual component of religion seemed negligible: paradoxically, the existence of obligatory practices (such as rites) pertained to morality more than it did to religion. The definition of the relationship between religion and morality was also problematic in Durkheim’s early works.
6From the link thus established between obligation and religion there followed, however that may be, a univocal determination of the social function of the latter (b):
“[L]aw and morality aim to maintain a balance within society, to adapt the latter to the conditions which surround it. Such must be the role of religion. If it belongs with sociology, it is because it exerts this regulatory influence on societies.”
8Religion, said Durkheim, was a “collective discipline”, “a school of disinterestedness and self-denial” (1886a, 196, 307). It therefore had more to do with the way society regulated itself and its members, than with the way individuals were attached to society. In the vocabulary of the Durkheimian theory of social bonds, we could say that, at first, religion for the young Durkheim was a factor for social regulation rather than integration.
9The question of the future of religion (d) remained open. Critical of explanations based on the idea of evolution (Wundt), Durkheim was also skeptical about the disappearance of religion from modern societies. And yet, if we accept the definition of religion as a “common faith”, a “belief one is subjected to without discussion”, it is hard not to think that in modern societies, where regulation is more and more structured in legal-rational forms, religion is not condemned to an irreversible decline. It is precisely on this last point that The Division of Labor in Society introduced new ideas. 
10In his thesis, Durkheim spoke of religion uniquely in relation to mechanical solidarity,  the weakening of religion being, in his view, proof of the “progressive indetermination” of the collective consciousness.  The contraction of this consciousness was accompanied by the evolution of religious ideas (which was thus, in reality, a “regression” or “contraction” [1902b , 144]): the objects of religious beliefs became more and more abstract and transcendent,  while the individual variations became more and more marked and numerous. The “religion of the individual” itself, which for Durkheim represented the modern form of religiosity, was certainly a “common faith” (i.e. a religion according to his definition at the time) but it differed from traditional religions and, in particular, it “attaches us to ourselves”, that is, it “does not constitute a real social bond” (Ibid.: 147, our emphasis), which is contrary to what he would argue later. We would therefore be in the presence of a shared belief which, nevertheless, did not play a real social role: the loss of this function was the price religion had to pay for the advent of modern forms of solidarity.
2 – Suicide and Early Studies in the Sociology of Religion (1897-1899)
11After a brief survey in his Rules of Sociological Method (see in particular 1947 , 62n., 105-106, 122), Durkheim went back to the subject of religion in Suicide and his attitude was no longer the same. He approached religion in chapters 2 and 3 of Book ii dealing with egoistic suicide. The high suicide rate among Protestants, stated Durkheim, depended on the development of free inquiry which depressed the “collective credo.” Consequently, in Protestant societies religion was not able to “act prophylactically” against the tendency toward voluntary death (1897a, 172). As egoistic suicide was defined by a lack of social cohesion, it followed that, if cohesion decreased just as the collective credo weakened, then the social function of religion (b) resided in its contribution to the integration rather than the regulation of societies. This thesis was the opposite of the one previously argued by Durkheim. That this was now his position was confirmed by the fact that Durkheim did not mention the decline of religious beliefs to explain the increase in the number of anomic suicides due to a lack of rules. It would therefore seem that Durkheim had performed a clear U-turn on the subject of the social function of religion. 
12Durkheim’s interest in religion continued in the first chapter of Book iii: “The Social Element of Suicide.” His break with previous texts now concerned the question of definition (a). Having reiterated his critique of the psychological explanations of the religious phenomenon, Durkheim wrote:
“The power which thus commanded respect [by the individual] and became the object of his adoration was society, of which the Gods were but its hypostasized form. Religion is, in fact, a system of symbols by means of which society becomes aware of itself; it is the way of thinking characteristic of the collective being.”
14The novelty of these lines on the subject of religion compared to the previous passages which had spoken of religion as a “common faith”  can be summarized in the following way: religion was defined on the basis of the relationship society had with itself and not on the basis of the relationship between the individual and society; it consisted of a way for society to become aware of itself, by a self-representation; this self-representation cannot be false or illusory because it concerned the way society thought; this way of thinking involved the use of symbols. The emphasis was still laid on the modality of beliefs, but Durkheim no longer turned his attention to their obligatory character.
15From this point on, the main difficulty that the Durkheimian sociology of religion would come up against was of a cognitive nature: why did society have to become aware of itself and, in particular, why did this happen precisely in a symbolic-religious form? Implicit from 1897, these questions would find their answer several years later and in particular beginning with the 1906-1907 course on religion.
16Finally, while nothing had changed in Suicide as far as the relationship between religion and morality (c) was concerned – as this remained founded on the element of obligation shared by the two areas (1897a, 381)  – Durkheim saw the future of religion (d) differently than in his Division of Labor in Society (Ibid.: 378-384). The religion of the individual was now presented as an aspect of the process of social diversification which led to organic solidarity and, thus, as a phenomenon able to give rise to social bonds:
“This worship of the human being […] [,] [f]ar from detaching individuals from society and any goal which surpasses them, […] it unites them within the same thought and makes them serve the same work”.
18Incidentally, the sacrality of the person as an element of social cohesion would lie at the heart of Durkheim’s intervention in the Dreyfus Affair (1898b, 267-268).
19The passage in Suicide on the definition of religion as a system of symbols also stands out because in it Durkheim made a footnote reference to The Golden Bough (1890, 2nd edition 1900) by James Frazer – an author destined to become a major ethnographic source for him, even when they did not agree  – and because it contained the first occurrence of the word “totemism” in Durkheim’s œuvre (1897a, 353). At the same time, Durkheim began to underline the explanatory value of religious phenomena: “In principle, everything is religious…” (1897e, 253). Is it then around 1897 that we should situate the first real turning point in the Durkheimian sociology of religion?
20The answer seems to be “yes” according to the paper on “The Prohibition of Incest and Its Origins” (“La Prohibition de l’inceste et ses origines”) published in the first volume of L’Année sociologique (1898a). The text dealt with totemism as a form of religion and saw the totem as the symbol of a community. The definition of religion as a system of symbols by means of which society thought of itself (the definition found in Suicide) was therefore a generalization of the case of totemism. Durkheim explained the totemic derivation of the prohibition on incest in the following way: (1) The totem was the clan’s ancestor, that is, a given individual (not a species). (2a) The clan members had with it, and therefore with each other, a relationship of substantial identity or “consubstantiality.” Indeed, following the principle of sympathetic magic according to which a whole and its parts are equivalent (2b), the totem remained whole within the individuals (parts) that originated from it. They were made, literally (2c), of the same substance as it (no other substance circulated within the clan). They formed, between them and with it, “one body, ‘one flesh’, one blood” (Durkheim 1898a, 83) . (3a) The admission of a stranger into the group was tantamount to establishing a new kinship relationship. This took place by means of a blood covenant, a ritual practice through which the new member came into contact with the group’s blood, for example, through inoculation or ingestion (sacrificial meal). This showed that social belonging meant sharing in the same substance – a communion – and that this substance was precisely the blood which was ascribed this role thanks to the vital function it fulfilled?(3b). Since the totem was the blood as well as a god, it followed that the blood was a “divine thing.” Like all things divine, it was taboo. As with the blood, women too were taboo because “they spend part of their lives in the blood, so to speak” (Ibid.:?85). Hence the prohibition on incest and the rule on exogamy, that is, the matrimonial practice which required men to go find their wives outside their clan.
21It was from Frazer, as well as from Robertson Smith and Sidney Hartland, that Durkheim drew most of the arguments for this explanation, often with literal fidelity as the following table shows:
Sources of the Durkheimian explanation of the taboo on incest
Sources of the Durkheimian explanation of the taboo on incest
22Nevertheless, Durkheim distanced himself from his sources on the subject of the totem itself which for him, coherently with his thesis on the consubstantiality between the totem/ancestor and the clan members, was not “an animal or plant species, but a given specific individual” (1898a, 83, n. 1).  The substantial unity of the clan meant that the totemic kinship was different from the biological one. While the latter was regulated through descent and was structured by degrees, in the case of the totemic kinship belonging or being foreign to the clan were absolute.  It was precisely because the totemic substance was indivisible that admission to the clan could be achieved “in another way than by birth right” (Ibid.: 84) (i.e. by the blood covenant). The belief in a substantial identity was the sign of a social bond which did not draw on any “natural” fact but was entirely constructed at a symbolic level. Already for Durkheim, and well before Claude Lévi-Strauss, the prohibition on incest marked the passage from nature to culture.
23In the 1898 paper we were faced with a social relationship which stemmed directly and entirely from phenomena of a religious nature (question b). Supporting the definition of religion proposed in Suicide (the second Durkheimian definition of religion; question a), this text illustrated a case in which a moral norm aimed at disciplining sexual instincts emerged from religious beliefs which did not necessarily imply it (question c). Was this then the “sociological approach to the study of religion” which Durkheim said he discovered during that period thanks to the works by “Robertson Smith and his school” (1907a, 404)? He used the same expression in 1898, in a letter to Marcel Mauss, in reference to the authors he quoted in “The Prohibition of Incest” (1998a, 174). Nevertheless, the case studied in this paper dealt only with an archaic society, consisting of a “homogeneous and compact mass” (a society characterized by mechanical solidarity as discussed in The Division of Labor in Society). To what extent could the conclusions drawn by Durkheim be generalized?
24The correspondence between Durkheim and Mauss on the subject of sacrifice during the summer of 1898 shows how this point was controversial. For Smith, sacrifice was only a means for the faithful to renew the clan’s substantial identity by eating the totem. Mauss reproached Smith, as well as Frazer, for having exaggerated this communional function of the sacrifice: if religion made it possible to establish social relationships, this would not take place only through a “direct fusion of the human life and the divine life.”  Mauss (together with Hubert) preferred the idea of communication to that of communion: the sacrifice was a procedure aimed at establishing and regulating the communication between the sacred world and the profane world, by means of a thing which was first sacralized and then destroyed during the ceremony.  Thus, through his critique of communion, Mauss questioned a fundamental element of the text on incest. Durkheim initially answered by defending Smith and Frazer, but ended up admitting that their approach was unilateral. On the other hand, Mauss and Hubert understood the duality present within the sacrifice, whose function was to bridge the gap between two areas which would otherwise remain separate: “And this duality, Durkheim wrote to his nephew, is very important since it makes it possible to glimpse everything that comes from it.” 
25Durkheim drew his conclusions from this discussion in his study “Concerning the Definition of Religious Phenomena” (“De La Définition des phénomènes religieux”) published in L’Année sociologique in 1899. From Mauss he took and retained the idea that the duality was a constant feature of the religious domain: namely, the duality between the profane and the sacred, in which he now saw the expression “by means of a symbolic language of the duality between the individual and the social” (1899, 163). Thanks to his students’ work,  the sacred-profane conceptual pairing – which would play an increasingly important role in Durkheim’s theory of religion – entered his new definition of religion, the third one of his œuvre. To obligatory beliefs, Durkheim added two elements suggested by Hubert and Mauss in their essay on the subject of sacrifice, namely, rites (or practices – the terms are synonymous) and sacred things: 
“So-called religious phenomena consist of obligatory beliefs connected to definite practices that are related to specific objects within these beliefs.”
27With this definition, Durkheim established first and foremost what religious phenomena were not. What did not pertain to religion were the moral and legal norms – which did not require obligatory beliefs (this thesis had already been argued by Durkheim in the 1880s; question c) – or those “common secular beliefs” such as faith in progress, in democracy and, one might add, the religion of the individual, from which rites were absent (Ibid.: 157, 159). What Durkheim seemed to say here was that modern societies could give rise to new commandments and new dogmas but not to new religions (question d). Also, the introduction of the notion of sacred things allowed him to go further with his critique of the definitions of religion centered on the idea of god (question a). The novelty compared to his early texts was that he did not limit himself to rejecting this idea for methodological reasons; he also indicated a category of objects which were at the center of worship independently of it and from which it was derived.  The notion of divinity, far from being merely the result of an individual mental projection, addressed the need to organize “the confused mass of sacred things”: it was “a grouping and unifying principle” which could form under certain conditions, but without being either original or necessary (Ibid.: 153). According to Durkheim, sacred things were the mark of a diffuse power which did not need to become individualized in order to manifest itself. In short, what he sketched out here was a sort of sociological justification of atheism. 
28However, the operation begun in “Concerning the Definition of Religious Phenomena” remained, as Durkheim himself admitted, a “preliminary operation” (Ibid.: 140). In the end, the 1899 paper – also a transition text – raised more questions than it solved. Durkheim made in it the inventory of research still to be done:
“The forces before which a believer bows […] are the direct product of collective feelings which were led to take on a material dimension. What are these feelings, what social causes awakened them and made them express themselves in this or that form, what social goals does the organization thus formed seek to address? These are the questions which the science of religions must deal with…”
3 – Toward The Elementary Forms
30At the beginning of the century, Walter Baldwin Spencer’s and Francis J. Gillen’s study The Native Tribes of Central Australia (Spencer and Gillen 1899) opened a new phase in the discussion on totemism. Both Durkheim and Mauss felt that the work done by the two British ethnographers marked a turning point in the study of religious phenomena: for the first time, a totemic system had been observed directly, “in its unity and integrality.”  The study presented “an enormous mass of facts”, in particular with regard to rituals, but also a “very serious” observation from a Durkheimian point of view (Mauss 1900a: 205-206): in the totemism of the Arunta tribes – the only ones whom the authors had studied from a religious point of view – there was no ban on eating the totem and, most importantly, there was no rule on exogamy. This led to two possible conclusions: according to Spencer and Gillen, totemism was indeed a religion but not a principle of social organization; according to Frazer, on the contrary, given the absence of these taboos (dietary and sexual), totemism was only a magic-economic enterprise whereby each clan took on the responsibility of looking after a given natural resource (animal, plant, atmospheric) for the benefit of the whole tribe.  In neither case was there a relationship between religious phenomena and society.
31Durkheim’s response came in 1902 in his paper “On Totemism” (“Sur le totémisme”), in which he showed that the totemism of the Arunta was the late development of an initial totemism,  of which there were some traces left: in particular, the Arunta still had the dietary taboo and the rule on exogamy, albeit in a weakened form. This was enough for Durkheim to restate the religious nature of totemism, as well as the relationship between it and the matrimonial organization of Australian societies – a subject he would discuss again in a paper in 1905 (Durkheim 1905). Once this point was made, Durkheim was able to put Spencer’s and Gillen’s discoveries to good use, in particular their description of the intichiuma, the totemic group’s annual ceremony. This involved a complex set of rituals (songs, dances, prayers, etc.) characterized by a state of collective overexcitement, at the center of which lay the sacred things, and ending with the ritual consumption of the totemic animal or plant. The intichiuma was a “positive” rite, that is, it did not consist of bans, but involved “active performances” (1902a, 347) and even breaches of ordinary social norms. Durkheim and Mauss saw in this the example of the “totemic sacrament” they had been looking for and whose existence Smith had only “guessed.” 
32It was as though the work done by ethnographers in Australia  had produced no change in the Durkheimian theory of religion comparable to the one which had taken place after he had read the works by Smith and his school. In reality, Durkheim’s position underwent considerable transformation during that period. Obligatory beliefs were still the central element of the 1899 definition of religion (question a). The obligatory nature of beliefs, which was due to their social origin, was transmitted to certain objects (the sacred objects) by surrounding them with bans and regulating the practices which concerned them. After that date, the criterion of obligation began to give way to the sacred things. This was probably not unrelated to the discovery of a positive totemic rite, in which the sacred things were central and the bans were secondary. In his 1906-1907 course, the definition of religion – by now being formulated for the fourth time – no longer explicitly mentioned the obligatory nature of beliefs, only the sacred things: religion was “a system of beliefs and practices related to sacred things – beliefs and practices shared by a given group” (1907b, 70).
33Alongside the criterion of obligation, the distinction between religion and morality was also being brought into question (question?c). Until that point, Durkheim had based this distinction on the fact that the dimension of obligation did not pertain to the same objects in the two cases (beliefs and practices in the case of religion, practices only in the case of morality). Instead, in his 1906 speech at the French Philosophical Society (Société Française de Philosophie) on “The Determination of the Moral Fact” (“La Détermination du fait moral”), Durkheim highlighted what morality and religion had in common: they were both methods for constructing ideals which shared a relationship with the sacred and were distinct from each other only through “difference[s] of degree” (1906a, 69).
34While the criterion of obligation was fading away, the hierarchy of the sacred objects was changing. In his paper on incest, Durkheim had seen in blood the expression of the religious principle par excellence: “the blood eminently contains the common principle which is the soul of the group” (1898a, 84). However, three years later, in his paper on totemism, it was the totemic being that had become pre-eminent, namely, the member of an animal or plant species (e.g. the kangaroo): in order to renew the “material as well as mystical” substance through which they could become members of a clan, individuals took part in a ritual meal where they absorbed “several particles of the being that is supposed to possess this substance most eminently” (1902a, 345). In 1906-1907, the top place on the sacrality scale went to the totem symbols – such as the churinga: flat stones or pieces of wood carrying a drawing of the totem – that is, to the “sacred things” in the strict sense of the term:
“It is these simple images and not the beings they symbolize which are the eminent source of religiosity.”
36What then is the function of the sacred things? In the text on the definition of religious phenomena, in which the problem was posed for the first time (“we need to know what these sacred things consist of and how we can recognize them” 1899, 154), Durkheim limited himself to stating that collective feelings “were led to take on a material dimension” (Ibid.: 161). In the paper on totemism, the sacred things were an extension of the totemic species, they were “transformed totemic beings” (1902a, 349), a connecting point between the original worship of animals and plants and the worship of ancestors. It was not until the 1906-1907 course that the sacred things acquired complete functional autonomy, as signaled by the fact that “the images of totemic beings are more sacred than the totemic beings themselves” (1907b, 82).
37However, the centrality acquired by the sacred things in the Durkheimian conception of religious phenomena raised new problems. The first one was that, according to ethnographers, the churinga were sacred objects which marked out individuals: for Spencer and Gillen, as well as for Frazer, they were a sort of external soul of the individual and, thus, the expression of a strictly individual kind of totemism. The second one was that since 1904 Durkheim had believed, in keeping with one of Mauss’s arguments (Mauss 1909 ), that totemic worship was directed at an impersonal principle, a sort of diffuse power or force, which he referred to by using the Melanesian term mana; but how could this principle be reconciled with the sacred things, which were concrete, specific objects? Finally, it remains to be explained why the images and symbols of totemic beings were invested with such a specific sacrality.
38In order to solve these difficulties and finally answer the question he had raised in Suicide – namely, that of why society must become aware of itself in a symbolic-religious form – Durkheim outlined in his 1906-1907 course a complex explanation of the genesis and function of sacred things.  Since it was in rites that the importance of sacred things became evident (1907b, 94), this was where one had to start, in particular the positive rite of the intichiuma. Here was Durkheim’s interpretation: Each year, the group, which normally lived dispersed on the territory, would come together to perform this great ceremony. From this coming together there emerged an intense, passionate excitement which, like any strong emotion, materialized in the concrete form of the object which caused it or, “if this object is not easily representable, [then] in its sign, its symbol” (Ibid.: 97). The idea of the sacred merely expressed this collective experience of a world which was radically different from that of ordinary or profane existence. The symbols of a clan, being at the center of this experience, became sacred in their own turn and even became the sacred objects par excellence. 
39We can see how through this reasoning Durkheim tried to solve the problems he had come across. First of all, it was only as the symbol or emblem of a clan that a certain object became sacred: in other words, there was no individual totemism without collective totemism. Then, Durkheim overcame the possible contradiction between sacred things and the impersonal principle of mana by seeing in these elements the two poles of a process punctuated by the rhythm of social life: the force superior to individuals (mana) which was unleashed by the ritual coming together would enter the sacred objects, where it would be preserved during the long profane season, thereby maintaining the group’s unity. As Mauss had suggested in 1899, the sacred-profane pairing did not express a static opposition between society and the individual; rather, it constituted a dynamic polarity on which the establishment and maintenance of social bonds depended (question b). They depended on it and perhaps they would continue to depend on it in the future, Durkheim suggested, even though the development of science and morality would probably impose profound changes on religion (Ibid.: 98; question d). 
40In short, at the beginning of the twentieth century, the Durkheimian analysis of religion was enriched by the study of rites and sacred things, which were added to the composition of religious phenomena alongside beliefs. In his effort to connect these three aspects in a coherent whole, Durkheim came to see in the polarity between the sacred and the profane a factor for structuring social relationships, rather than a simple reflection of the opposition between individual and society. We might therefore wonder whether a new theoretical change in Durkheim’s view of religion did not reflect the work carried out during those years. The four-question schema will help guide us one last time.
41Concerning question (a), Durkheim, who had started from a definition of religion centered on the modality of belief (the common faith), defended until 1899 the importance of obligation as a distinctive mark of religious beliefs. At the same time, in Suicide he started to see in religion a system of symbols which fulfilled a social function. It was only around 1906/1907 that this social function seemed to be understood, the sacred things now taking the place of obligation in the definition of religion.
42Concerning question (b), Durkheim, who had started by limiting the social function of religion to a regulatory role, took increasingly into account the integration which resulted from religious beliefs and practices and which he saw as a form of communion. After 1899, this view was in turn questioned. What Durkheim now saw at play in religious phenomena were complex processes relevant to both dimensions of a social bond: integration – of which communion was just one modality – and regulation. For example, the intichiuma ritual showed a moment of deregulation (relaxation of social norms) accompanied by very strong integration and followed by the reestablishment of social norms and relations of average intensity. 
43As far as question (c) was concerned, for a long time Durkheim searched for the distinction between morality and religion in the difference between the objects that obligation pertained to in these two cases: beliefs and practices in the case of religion, practices only in the case of morality. As the dimension of obligation lost its importance in the definition of religion, it became less imperative to establish the criterion according to which religion and morality should be separated. The study of the genesis of the prohibition on incest persuaded Durkheim that the passage from religion to morality, where it did take place, could be analyzed as a historical fact without introducing the idea of a necessary transition from an inferior to a superior level. He therefore tended to conceive of religion and morality as two autonomous areas sharing certain characteristics thanks to their common social origin.
44As for question (d), the new definition of religion based on the opposition between the sacred and the profane allowed Durkheim, after a long hesitation, to think differently about the future of religion: if it was true that the sacred-profane duality constituted an essential component of social reality, modern societies would not be characterized by any disappearance of religion. What we would witness would be a transformation of religion, that is, a continuous redefinition of what was sacred and what was profane.
45The following table summarizes the evolution of Durkheim’s position with respect to the four questions of his research program in the sociology of religion.
The stages of the Durkheimian sociology of religion: 1886-1907
The stages of the Durkheimian sociology of religion: 1886-1907
46Thus, the final years of the nineteenth century marked a real turning point in the history of Durkheim’s sociology. Suicide and the 1898 paper on incest showed a profound change in his view of religion, which has to be attributed to the influence of Smith’s school, as Durkheim himself suggested. However, the theoretical framework which emerged during the years after the “revelation” of 1895 was not definitive and was still far from anticipating the arguments put forward in The Elementary Forms. On the contrary, the debate animating the beginning of the new century, within and without the Durkheimian camp, led Durkheim to once again change his mind about these four fundamental questions: around 1907 there occurred a second turning point in his theory of religion. This was then an extraordinary and composite intellectual journey made up of continuities, innovations and wanderings, which led Durkheim from his reviews of works by Spencer, Wundt and Guyau to his last book. The Elementary Forms represented the conclusion of this journey. It was certainly not its logical destination implied at the beginning. It was a great result, but not the only one possible.
See Pickering 1984, 47-91; 1993; Wallwork 1985.
See Jones 1981 and 1986.
Nevertheless, Durkheim’s view on the other three questions remained the same: (a) religion was defined by the modality of belief since the religious character of a feeling rose to the surface every time “a slightly strong conviction [was] shared”, that is, each time we were faced with a “common faith” (1902b , 143, 147); (b) the main social function was regulation: religion “exerts constant constraint on individuals”, “religious life is made up entirely of self-denial and disinterestedness” (Ibid.: 59-60); (c) religion was distinct from morality since it concerned ideas rather than practices (Ibid.: 144).
His sources were the works by Herbert Spencer, John Lubbock and Albert Réville.
On this subject, the most important sections are in book I, chapter?V, § 5 (1902b , 142-148) and book II, chapter II, § 1 (272-276).
According to the succession from fetishism to animism, to polytheism, and then Christianity (we note the absence of totemism), 1902b , 273-274.
In the conclusion (1897a, 432), Durkheim suggested that the change in the role played by religion (from regulation to integration) coincided with the transition from archaic societies to modern societies.
This definition was taken up again in book I of Suicide (1897a, 173).
In his course on Moral Education (1898-1899, in particular lessons i and vii), which closely followed Suicide, the relationship between religion and morality was seen as a form of subordination: religion expressed symbolically (that is, irrationally) a moral content which needed to be “expressed by means of a rational language” (1963 [1898-1899], 7), that is, a language which was based on empirical realities, was not static (was open to the change of moral ideals over time), and was coherent with regard to the essential element of morality that was autonomy. In other words, in Moral Education religion was only a social phenomenon subsidiary to morality. This introduced a certain theoretical tension with Durkheim’s reflections on religion as expressed in Suicide.
Durkheim would later become a strong critic of Frazer. We might suppose that he discovered his works late and certainly not before 1894-1895.
Durkheim wrote “one flesh” (une seule viande) between inverted commas following Frazer’s indication, who used these words to translate a native expression (1887, 59; see table below).
However, see Frazer 1887, 1-2: “A totem is a class of material objects […]. As distinguished from a fetish, a totem is never an isolated individual;” and Smith 1889, iii, 124: “some natural kind of animate or inanimate things.”
See Smith 1885, 52-53: “To us, who live under quite modern circumstances and have lost the tribal idea altogether, kinship is always a variable and measurable quantity. We have a strong sense of kindred duty towards parents or children, not quite so strong a one towards brothers, and a sense much less strong towards first cousins […]. Nothing can be clearer than that the original doctrine of kinship recognized no difference of degree.”
Hubert and Mauss 1899, 39 on blood covenants.
In Mauss’s language, the moment of socialization (of the thing or the victim, of the officiants themselves) would always have to be followed by a moment of desacralization; without it, he wrote to his uncle, “individuals or society sink into fatal mysticism. The sacrifice has to be a passage, a moment, one of the various links in an individual’s social life. It must not be all of it, otherwise it absorbs and annihilates it; it thus also fails in its social and moral function” (1998a: 159-160, July 1898).
Ibid.: 174, August 1898.
See Isambert 1976.
“Thing” here means, in its strict sense, an individual nonliving material entity: a piece of wood, for example, a stone, or a rock.
To make his point, Durkheim relied on the standard definition of the totem as an animal or plant species (1899, 151), a definition which he himself had questioned in his paper on incest on the basis of the conviction that the totem was a group’s ancestor.
Durkheim 1899,153: “We can now better understand how there can be atheistic religions such as Buddhism and Jainism.”
Durkheim 1902a, 315. A second work on the tribes of Central Australia would follow in 1904 (Spencer and Gillen 1904; see Mauss 1905). See Testart 1998; Zerilli 2001; Rosa 2003, 91-103, as well as William Watts Miller’s contribution in this volume.
Frazer 1899, see Mauss 1900b.
This hypothesis had already been formulated by Mauss 1900a, 214.
Mauss 1900b: 218; see Durkheim 1902a, 320, 347.
Durkheim continued to follow this work attentively: see in particular his reviews of works by Howitt and Strehlow in L’Année sociologique (Durkheim 1906b, 1910; see also the critique of Lang 1907c).
The explanation was taken up again and developed in chapter vii, book ii of The Elementary Forms of The Religious Life.
Thus, Durkheim went from the idea that the totem was the emblem of the clan to the more original one that society needed emblems or symbols in order to exist: he went from the emblematic function (fulfilled by beings which were not emblems) to the emblematic form per se. A possible source of this development of his view of totemism is the work of the Hungarian philosopher and psychologist Julius Pikler Der Ursprung des Totemismus (Pikler and Somlò 1900), which Durkheim probably came across in 1905 (Némedi 2008).
Durkheim made this point in 1908 in his speech at the French Philosophical Society, in reference to Émile Boutroux’s Science and Religion (Durkheim 1909 ).
See Steiner 2000.