1
The One and the Many

The "Anthropology of Islam" in a Muslim Community

This book is about Islam in a single neighborhood of a relatively large town in the West African country of Côte d'Ivoire. As a description of a small community in a corner of the globe remote from North America and Western Europe, it is typical of the work that anthropologists tend to undertake. Admittedly, the ethnography of small communities—communities non-anthropologists might even be tempted to label "insignificant"—might seem a curious approach to the study of a phenomenon of such global spread and significance as Islam. Yet it is the premise of this book that analysis of the religious beliefs and practices of Muslims in the neighborhood of Koko constitutes a reasonable, valid, and significant way of contributing to the understanding of the religion of Islam. Of course, that premise underlies, not only this work, but the whole of the burgeoning field of the "anthropology of Islam."

The "anthropology of Islam," as a deliberately constituted field of academic inquiry, has only emerged within roughly the past twenty years. Its birth was heralded in 1968 by the publication of Clifford Geertz's book Islam Observed. Of course, anthropologists had not previously ignored Muslim societies, much less the importance of religion in those societies. E. E. Evans-Pritchard's study of the Sanusi of Libya (1949) and Geertz's monograph of religion in a Javanese town (1960) are among the most prominent, although hardly isolated, examples of prior anthropological interest in Islam. Yet even the titles of these two books—The Sanusi of Cyrenaica and The Religion of Java— are revealing. Both titles end with the

name of a specific locale, yet neither refers to Islam by name, but only obliquely: the Sanusi, a Sufi order, are ipso facto Islamic; and the majority of Javanese do claim, in one way or another, to be Muslim.

As their titles indicate, such works fall within a well-established monographic tradition. Anthropologists were expected to study specific "cultures" or "societies" situated in some precise, and usually exotic, corner of the globe. "Religion" in one form or another was conceived to be an essential component of such a culture or society. If some or all of the members of this culture happened to be Muslim, it was likely that the anthropologist would have something to say about Islam in that particular locality. Indeed, such a discussion might be essential to any comprehensive description. Thus, for example, Horace Miner's The Primitive City of Timbuctoo (1953) devotes an entire chapter to Islam, sandwiched between chapters on "Elementary Economics" and "Genii and Witches." A monograph might even focus quite specifically on the religion of a particular culture, as Geertz's 1960 book does on that of Java, or, more precisely, of the town of Pare, alias "Modjokuto." Geertz had to concern himself with Islam in it because it was, to one extent or another, "the religion of Java."

In short, until relatively recently, anthropologists did not set out to study Islam per se, but rather the religion of some particular culture, society, or locality. To a certain extent, there were informal biases against studying Islam, and even Muslim societies. In the first place, the "anthropology of religion" had, since its origins in the nineteenth and early twentieth centuries in the writings of Spencer, Tylor, Frazer, and Durkheim, been concerned first and foremost with "primitive" religions, or, as they are now more euphemistically labeled, "traditional," "nonliterate," or "nonscriptural" religions. The nature of religious texts posed another problem for anthropologists who might choose to study Muslim societies, particularly outside the Arabic-speaking world. Anthropologists were usually too busy acquiring minimal competence in the local vernacular to have time to learn clas-

sical Arabic, the language in which Islamic texts are written. The study of these texts, and consequently of Islam, was by and large left to "Orientalists," whose expertise lay precisely in their exegesis.

Such reticence about studying Muslims and, a fortiori, Islam, notably in parts of the world such as sub-Saharan Africa where both anthropologists and Muslims might be found in relative abundance, was not, as we have seen, an absolute barrier to the production of books, chapters, and articles by anthropologists dealing with Islam and things Islamic. Nevertheless, it was only with the publication of Geertz's Islam Observed that Islam in and of itself became an explicit object of anthropological study. The central problem raised by the book, and indeed by the "anthropology of Islam" as an academic discourse, is the diversity of religious beliefs and practices in the Islamic world. Geertz approaches this problem by contrasting two national Islamic traditions—Indonesian and Moroccan—at the symbolic antipodes of the Muslim world. This explicitly comparative perspective also characterizes subsequent books such as Ernest Gellner's Muslim Society (1981) and Michael Gilsenan's Recognizing Islam (1982), as well as various theoretical articles surveying the field of the "anthropology of Islam" (e.g., Zein 1977, Asad 1985). As one might expect among anthropologists—a fractious lot, overall—there is not a great deal of obvious consensus among these various authors about how exactly one ought to go about the business of developing an "anthropology of Islam." Yet their very real theoretical differences have obscured the extent to which they are all engaged in a common enterprise and grappling with a common set of questions that had not previously troubled the community of anthropologists, even those who studied Muslim societies and who wrote in one way or another about religion.

Perhaps the most fundamental assumption these authors share is the conviction that the "anthropology of Islam" is, in itself, a meaningful enterprise. This assumption is not entirely self-evident; it presumes the reality of "Islam." These writers postulate that "Islam" is more than simply a label for

a variety of phenomena that have little, if anything, intrinsically in common—in other words, that the concept of "Islam" is not, epistemologically speaking, analogous to, say, the concept of "totemism." In the nineteenth century, and for part of the twentieth, there was widespread academic consensus that certain kinds of beliefs and practices constituted "totemism," and that examples of "totemism," if not "totemic religion," could be found in Australia, Polynesia, North America, and Africa among other places. In retrospect, it is clear that "totemism" was really an invention of anthropologists, an amalgam of unrelated traits that tended to occur separately more often than together. It was an artifact of academic discourse rather than of the exotic cultures the anthropologists purported to describe.^[1] Obviously Islam, unlike "totemism," is not an invention of Western academics. Real people all over the world freely identify themselves as Muslims; few, I daresay call themselves "totemists."^[2] Even so, such self-identification does not in itself justify the analytical usefulness of the label. People have also identified themselves as "members of the Aryan race," a fact that would hardly in and of itself justify an "anthropology of Aryans."

Admittedly, anthropologists have not, by and large, challenged the reality of Islam, as they have challenged the reality of "totemism" and of "the Aryan race." But, until recently, Islam was simply taken for granted, precisely because it was not itself an object of study. For example, Islam could be conceptualized as a reality initially external to the culture an anthropologist was studying, as in Joseph Greenberg's monograph on The Influence of Islam on a Sudanese Religion (1946). Seen in this way, Islam would appear to be an externally constituted set of beliefs interacting with some other, internally constituted, set of beliefs to produce a syncretic synthesis. It was the task of the "Islamicist" to describe Islam, and the task of the anthropologist to describe its "influence" at the "periphery" of the Muslim world, if not its local peculiarities within the Muslim "core."

Paradoxically, it is only with the emergence of an "anthropology of Islam" that anthropologists have no longer been

able to take "Islam" for granted. The problem that has emerged is, how can the very diverse—if not diverging—religious beliefs and practices of Muslims be comprehended within a single idea of "Islam"? Perhaps the most obvious solution to the question would be to posit the existence of, not one, but multiple "Islams."^[3] Anthropologists sometimes seem to lend credence to this idea by using national or ethnic qualifiers to write about local Islamic beliefs and practices, implying by their use of terms that there exist both an "Indonesian Islam" and a "Moroccan Islam," a "Dyula Islam" and a "Hausa Islam."^[4] Such a formulation is theologically unacceptable to most Muslims, who assert that there is, and can only be, one Islam. This assertion is by no means a sign of naïveté. Muslims are as aware as Western academics of the diversity of beliefs and practices within their own religion. In the first place, the idea of a single and unitary Islam can nevertheless leave conceptual room for variability, exemplified for example by the notion of the four Sunni madhhab , or "schools" of jurisprudence. In the second place, Muslims may hold that particular groups and individuals, whether in error or through malice, label as "Islamic" certain beliefs and practices that are inconsistent with the one true Islam. Of course, Muslims disagree about what is or is not acceptable, about what is or is not Islamic, but this very disagreement assumes the existence of a single true Islam. For anthropologists to assert the existence of multiple Islams is, in essence, to make a theological claim, one most Muslims would not only deny but, they rightfully argue, anthropologists have no business making.

In any case, the empirical diversity of the religious beliefs and practices of Muslims is not only of significance to anthropologists. Aside from Muslims themselves—who, after all, are most directly concerned—this diversity has not escaped the notice of historians, Islamicists, and other academics interested in one way or another in Muslims around the world. However, anthropologists, precisely because their discipline is rooted in the ethnography of small-scale local communities, tend to approach the problem of diversity somewhat

differently from other scholars. Typically, anthropologists are inclined to interpret the phenomena they study as quasiorganic products of the particular, if not peculiar, features of a specific locality. Seen in this light, the "religion" of a community may be analyzed as an integral component of its overall "culture," or as a reflection of its underlying network of social relationships. When the religion in question is a socalled "traditional," nonscriptural religion, such an "organic" approach to its analysis seems less self-evidently problematic.^[5] However, Islam is obviously not a "product" of any specific local community, but rather a global entity in itself. The problem for anthropologists is to find a framework in which to analyze the relationship between this single, global entity, Islam, and the multiple entities that are the religious beliefs and practices of Muslims in specific communities at specific moments in history.

Certainly the easiest, but perhaps the least satisfactory, way to resolve this tension between the local and global aspects of Islam as practiced is to posit a neat theoretical dichotomy between a universal Islam on one hand and local culture or society on the other. Seen in this light, local practices would in fact constitute some sort of syncretic synthesis between an organically constituted "pre-Islamic" culture and a coherent, unitary, and preestablished Islamic faith. Alternatively, one might attempt to distinguish between constant and variable components of Islamic belief and practice in different communities. Constant features would constitute the essential "core" of Islam, whereas variable features could be explained in terms of local social and cultural peculiarities.

The "anthropology of Islam" has emerged out of a common refusal to accept such solutions, however acrimoniously its practitioners may debate one another on other grounds. In the first place, Islam does not exist apart from the specific beliefs and practices of diverse individuals in particular communities at precise moments in historical time. This undoubtedly seemed less self-evident to earlier anthropologists who studied Muslims on the "periphery" of the Islamic world, notably in sub-Saharan Africa. However, as anthropologists

turned increasingly to the study of Islam in its putative Middle Eastern "core," the range and nature of variability simply refused to melt away. In short, there was simply no place on earth where one could observe "pure" Islamic practice divorced from local "syncretic" accretions or deviations of one form or another. It was equally clear that, despite tremendous variability, Islam as practiced could not be reduced to a virtually infinite series of purely local idiosyncrasies. Practices initially judged "atypical" of Islam, and consequently deemed to be products of local culture or society, often turned out on closer inspection to have a far broader distribution within the Muslim world than initially imagined.

From this dilemma, the "anthropology of Islam" has carved itself out a specific theoretical space, between the particularities of the specific local communities anthropologists study intensively and the global features of a universalizing religious discourse, Islam. Explicitly comparative studies, such as those of Geertz, Gellner, and Gilsenan, represent one way of mapping out this space. However, even such explicitly comparative works are ultimately based on ethnographic fieldwork in specific locations, and so monographic studies of Islam in specific communities are now, virtually of necessity, part of an essentially comparative enterprise, an enterprise that seeks to reconcile, analytically rather than theologically, the one universal Islam with the multiplicity of religious ideas and practices in the Muslim world.

In any case, this multiplicity is not a feature of Islam that emerges only through the comparison of one local community with another. On the contrary, Muslims—not only clerics, but ordinary believers—are often acutely aware of alternative ideas and practices among other Muslims, either within their own communities or outside. This was precisely the case in Koko, the neighborhood that constitutes the subject of this book. It was not simply that individual Muslims were conscious of other ways of thinking or acting. More precisely, individuals defined their own religious practices with explicit reference to the religious practices of others. In the first place, Muslims, a minority in the region, contrasted

Islam to the religion of their "pagan" neighbors. More recently, within the context of the colony, and later independent nation, of Côte d'Ivoire, Christianity, the religion of the former French rulers, and nowadays of a sizable number of Africans, also furnishes an explicit point of reference. By and large, these contrasts with other, non-Muslim religions tend to be made complacently enough. For committed Muslims, differences between their own religion and the religions of others constitute a series of grounds for asserting the superiority of Islam over its immediate rivals.

However, the variability of religious ideas and practices in Koko was by no means limited to the contrast between Islam on the one hand and "paganism" and Christianity on the other. Rather, the beliefs of Muslims were just as often, and perhaps more saliently, contrasted with those of other Muslims. In the first instance, the practices of Muslims nowadays were contrasted with those of the past. Muslims in Koko were acutely aware that the practice of Islam in their neighborhood had changed in fundamental respects within the past fifty years. Certain quite specific features had been abandoned, and others had been adopted. It was universally acknowledged, at least in public discourse, that these changes were for the better. Thus the contrast between the present practice of Islam and its past practice—but only in certain, highly specific respects—was ultimately reassuring, much as was the contrast between Islam and other religions practiced in town.

There existed yet a third, and more troubling, sort of contrast, the contrast between discrepant notions about whether specific ideas and practices are or are not fully consistent with Islam. The changes that had occurred some fifty years ago were the product of just such a controversy, but since then other terrains of disagreement have taken their place. Muslims may rest assured about the superiority of Islam over other religions; they may have no doubts that they know better than to repeat the religious errors of their ancestors (or perhaps only of the ancestors of their next-door neighbors). However, Muslims can never be quite so assured that their own conception of Islam is correct, and that challenging

views are in error. Koko was—and remains—a terrain where discrepant conceptions of Islam confronted one another in competition for the allegiance of the Muslim community. Muslims in Koko must ultimately decide to commit themselves to one point of view or another. Such challenges do not, by any means, always take the form of militant confrontation, of factional strife. Nonetheless, the existence of alternatives remains present in the consciousness of Muslims, even if these alternatives are largely identified as the practices of "other" Muslims in "other" communities, for in any case some of these "other" communities can generally be found on the other side of town.

What I wish to suggest is that, at one level at least, the multiplicity of religious beliefs and practices emerges only in the context of salient oppositions: Islam is opposed to non-Muslim religions; the religious errors of the past to the practices of the present; one's own conceptions of what is or is not proper Muslim practice to the conceptions of others who explicitly disagree, or who may, more quietly, seem to doubt. In a sense, multiple conceptions of Islam are defined, not so much by consensus among their adherents, as by the cleavages that distinguish them from other recognized and rejected alternatives. Muslims, I would suggest, are acutely conscious of the differences between their beliefs and practices and the beliefs and practices of others, Muslim or not, past or present—perhaps even real or imagined.

In other words, the empirical variability of Islam is not simply an analytical problem for anthropologists, historians, or other academics concerned with comparing Islamic beliefs and practices in different places and/or times. Most Muslims, I strongly suspect, are aware of the existence of different conceptions of Islam from the ones they hold, of alternative ideas and practices that also lay claim to the name of Islam, but that are, to some degree, if not radically, inconsistent with their own. At the center of these conceptions are questions about what it means to "be Muslim"—in other words, about what ideas and, especially, what practices are acceptable, desirable, or obligatory, or else objectionable, if not prohibited. Conceptions of Islam, as opposed to one another in

specific places and times, revolve around disagreements over the status of specific practices, about the ways in which Muslims ought to act and the religious significance of different forms of action.

It might seem reasonable to assimilate these different conceptions of Islam to so many different "interpretations" of Islam. After all, Islam is not only a scriptural religion but, compared with other scriptural religions, a highly textual one. Answers to these questions are to be found in the texts; indeed, that is the very purpose of many, if not most, of these texts. However, like all texts, they are not free of ambiguities, and so they can be read in different ways. Western scholars (and probably many non-Western scholars) are inclined to explain different doctrines and practices in terms of different "readings" of the texts. However, this is not at all how Muslims in Koko explain such differences or, more critically, conceive of their own beliefs and practices in the light of potentially contradictory ones. For example, the standard textbook of Maliki fiqh , the Risala of Ibn Abi Zayd al-Qayrawani (1968:173), states that the minimum marriage portion must be one-fourth of a dinar. The Muslims of Koko and surrounding communities are aware of and respect this stipulation, so that one of the prestations involved in any marriage transaction is the robon dinari , the fourth of a dinar. The problem is that dinars have never been used in the region as a unit of currency. This is not in itself insurmountable, as long as some standard equivalent—in gold, cowrie shells, colonial or post-colonial currency—is socially recognized. Indeed, when I was inquiring in the field about marriage practices in 1973, I easily found such standardized equivalents. However, while it was certainly not the case that each particular village had its own unique equivalent, it became obvious that such equivalents were subject to local variation. The point is that these specific equivalences, and the fact that they might in fact vary, were never conceived as specific "interpretations" of the general rule. Indeed, to speak of an "interpretation" in the first place is to recognize, at least tacitly, that the text in question is ambivalent, that it can in fact be "read" in different ways,

even if only one of these readings is deemed to be "correct." On the contrary, informants asserted that the rule was perfectly straightforward: a valid marriage had to include the payment of one-fourth of a dinar. Individual informants not only stated that this was the case in their own community, and in every other Muslim community in the area, but even went so far as to assert that the standard equivalents were everywhere the same. In other words, it was flatly denied on principle that the text might be "interpreted" in more than one way.

There are a number of objections that might be raised to this example. The first is that the issue of the robon dinari is a trivial one. It is, however, extremely dangerous for outside observers to make a priori assumptions about which particular issues are or are not trivial. The very fact that equivalents did exist in every community suggests, on the contrary, that Muslims in these communities took great pains to observe this particular stricture as scrupulously and conscientiously as possible. A more powerful objection would be that informants were not always answering in perfectly good faith. Despite high rates of endogamy, not only within villages but also within descent groups, marriages between members of different villages certainly occur nowadays, as they did in the past. Those discrepancies in rates of equivalence that did not escape my notice must, from time to time, have come to the attention of local Muslims. To assert that the rule could be applied one and only one way was a convenient legal fiction, rather than a statement of fact. Even if this is the case, the fact that a legal fiction is necessary to protect the notion that texts can have one and only one interpretation is itself significant. This is apparent in the way in which Muslims spoke about openly recognized differences of opinion. Such differences included outright doctrinal controversy, for example in disputes between the majority of the Muslims of Koko and adherents of the so-called "Wahhabi" movement. On the other hand, Muslims who share the same overall doctrinal perspective may yet disagree about whether certain acts in specific contexts remain religious obligations, or

whether their performance is in fact an ostentatious pretense of piety. In all such cases, when it was clear that there were different opinions about specific issues, not to mention radically different conceptions of Islam as a religion, no one ever suggested to me that there might be two interpretations of specific texts, much less of "Islam" as a whole. On the contrary, different viewpoints were always characterized as "ignorance." Disagreement never implied the logical, possibility of an alternate (if misguided) "reading" of the same texts, but rather the notion that one party to the dispute (one's opponent) possessed an incomplete knowledge of otherwise unambiguous rules.

The fact that Muslims in Koko may contrast their own conceptions of Islam to those of others in terms of "knowledge" versus "ignorance" might seem to suggest that such conceptions are doctrinally coherent systems of Muslim religious thought, systems that may be more or less perfectly known by the individuals who subscribe to them. Seen in this light, the aim of the investigator—anthropologist, historian, Islamicist, or whatever—would be to reveal, as fully as possible, the component elements of such a system and their interrelationship. The most obvious way to undertake such an enterprise would be to focus on the teachings and writings of those individual adherents of any conceptions acknowledged as the most "knowledgeable"—prominent 'ulama' and Sufi shaykhs , for example. Such a strategy is more typical of Islamicists than of anthropologists, for whom such teachings and writings constitute only one element (albeit frequently, though not inevitably, an extremely important element) of broader systems of meaning. Unlike the discourse of the "knowledgeable," such systems are not consciously articulated by their followers, just as the rules of grammar may be totally unknown to the native speaker of a language. The role of the anthropologist, if one accepts the premises of such a semiological approach, is to uncover the hidden logic behind such systems of meaning.

Those who assume that the key to the system is to be found in the discourse of the most "knowledgeable" assume, at least implicitly, that specific conceptions of Islam are not

fully comprehended by the mass of believers, but only by those few who are able to articulate them. Conversely, semiologically minded anthropologists imply that all adherents comprehend the underlying system of meanings at one level, but that only academics such as themselves—generally outsiders—are capable of articulating it. Both perspectives, however, posit the existence of an internally consistent, logically coherent system of meaning, whether in the form of statements of doctrine or of shared, if tacit, understandings. I hardly wish to suggest that neither doctrine nor tacit understandings are relevant to the understanding of various conceptions of Islam, but I would definitely call into question the assumption that such conceptions necessarily entail a fully developed internal consistency. To the extent that such conceptions are most consciously articulated with respect to one another, Muslims, not only the learned but also ordinary believers, are able to express the salient points of difference between Islam as they understand and practice it and alternative practices or ideas. On the other hand, there may be considerable variability among ideas and practices within a single conception of Islam, and not only between differing conceptions. In other words, from the point of view of an outside observer, beliefs and practices within one single conception of Islam may appear inconsistent, if not contradictory. The anthropologist, after all, constructs a representation of a given conception of Islam out of a welter of individual statements, actions, and events observed and recorded in the field. The extent to which this representation reflects an underlying coherence is as much a testament to the anthropologist's ingenuity and his or her commitment to identifying such a scheme as it does the "nature" of that conception of Islam.

If the quest for coherence is, taken too far, illusory, how, then can one talk about, much less characterize, a specific "conception" of Islam? This book is an attempt to answer this question, taking as its center the conception of Islam held by the majority of Muslims in Koko in the 1970s and 1980s. The first part of the book is concerned with the ways in which different conceptions of Islam have emerged and confronted

one another in the community of Koko. These chapters attempt to situate conceptions in history, and consequently center on the themes of change as well as on controversy. However, I am not going to present a historical overview of Islam in Koko in conventional chronological order, with different chapters focusing on different periods of time. Rather, I wish to explore the themes of change and of confrontation between different conceptions of Islam from a variety of vantage points, which constitute the organizing themes of different chapters: for example, the sources of scholarly authority, or the centrality of different kinds of ritual. The second part of the book focuses on ideas and practices within a single conception of religion at a single moment in time. From the vantage point of an external observer, these ideas and practices are characterized by certain inconsistencies. For example, most Muslims in Koko state that it is highly meritorious to belong to a Sufi order, and yet, to the apparent despair of Koko's few Sufis, very few bother to join. In principle, the authority of local scholars is based on how much they know compared to their colleagues, yet it is next to impossible to formulate any clear system for ranking scholars in terms of who knows more than whom. Unlike those points of contention that serve to demarcate one conception of Islam from another, such inconsistencies, such points of ambiguity, are not of great concern to local Muslims. Rather, they appear inconsistent and ambiguous only to the extent that one attempts to fit a variety of observed statements and events into a single coherent, consistent system.

African Time, African Space

Within the "anthropology of Islam," the problem of multiplicity tends to pose itself in either the most global or the most localized terms. On one hand, Islamic ideas and practices in geographically and socially disparate localities can be compared and contrasted with one another; or, alternatively, analysis may focus on variability within the confines of a single community. However, any such opposition of "local"

and "global" space is obviously too crude; between a single neighborhood of a single town and the worldwide community of Muslim believers, there are a variety of "middle grounds" to be taken into account. Nations constitute perhaps the most obvious of these middle grounds, and indeed anthropologists have not hesitated to write about Islam in Morocco, Indonesia, and Iran, for example. However, such a national approach is clearly inappropriate to studying of Islam in sub-Saharan Africa in any historical depth. Nations like Côte d'Ivoire are very recent creations. Koko quarter is literally older than Côte d'Ivoire. In an African context, it makes more sense to speak in regional than in national terms. In other words, one can analyze Islam in Koko in the more general framework of Islam in West Africa as a whole.

Curiously, anthropologists have had very little to say about Islam in West Africa in general. By and large, the "anthropology of Islam" has bypassed sub-Saharan Africa,^[6] despite the large number of anthropologists who have studied African societies. Indeed, anthropologists have made noteworthy contributions to the study of Islam in specific West African communities,^[7] contributions that have exhibited a considerable knowledge of and sensitivity to issues in Islam. Nevertheless, the focus of these studies has been on issues other than the study of Islam per se, and anthropologists have by and large left the field of discourse on Islam in West Africa to specialists in other fields, for the most part Islamicists and historians.

However, in studying Islam within a specific local community such as Koko, it is essential to situate it within the broader context of Islam in West Africa. The way in which this broader context is characterized becomes crucial. In particular, change over time—the conversion of non-Muslims to Islam, the abandonment by Muslim communities of certain conceptions of Islam in favor of others—is hardly a phenomenon peculiar to Koko. The most obvious way of characterizing such changes in West Africa has been in terms of progressive stages of Islamization. Seen in this light, competing conceptions of Islam, locally as well as regionally, can be

identified as representing different stages. The problems with such an approach are legion. At its worst, this kind of developmentalism is easily combined with an explicitly or implicitly racist evolutionary discourse. Islamic civilization is contrasted to the intrinsic savageness of the generic African, who is only capable of assimilating Islam gradually, bit by bit. The following passage (published as late as 1959!) exemplifies this kind of thinking:

The difference between Islam as a developed civilization with a body of religious doctrine and the African religio-social systems with which it is in contact is so profound that the psychological shock of conversion would seem as great as with Christianity. It is not so in practice. The reason is that African Islam in contact with animists is characterized by a series of gradations which act as insulators passing Islamic radiation on, diminuendo, to animist societies. Thus the form in which Islam first makes its impact upon the animist seems little removed from animism. This gives Islam the advantage of rarely finding itself in direct contact with animists in a form whose cultural level is too high to render mutual understanding possible.

(Trimingham 1959: 33)

The author goes on to identify three stages "in the assimilation of Islamic culture": first, "the infiltration of elements of Islamic culture into animist life"; second, "conversion, characterized more by the break with the old order than the adoption of the new"; and, only finally, "the gradual process by which Islam changes the life of the community" (ibid.: 34). Different conceptions of Islam in West Africa are hierarchically evaluated in terms of their degree of Africanness; the less "African," the better. The purity (at least the relative purity) of Islamic civilization is directly contrasted to the essential animism of Africans.

Such crude characterizations, typical of some schools of colonial discourse about Islam in Africa, have fortunately and most justifiably fallen into disrepute.

However, there are still scholars who continue to conceptualize Islam in West Africa in terms of a series of progressive

stages of Islamic development. The leading exponent of such a view is Humphrey Fisher (1973), who, like Trimingham before him, has identified three stages of Islamization: "quarantine," where the presence of minority Muslim communities is tolerated, but these communities remain spatially and socially distinct from the majority; "mixing," where rulers declare themselves to be Muslims, but where religious practices are a syncretic amalgam of Islam and of traditional African religions; and finally "reform" in an attempt to purge Islam of such supposedly syncretic accretions. At one level, Fisher's stages can be taken to represent a chronological sequence that can be applied without much difficulty to West African history. The empire of Ghana, which flourished roughly between the eighth and eleventh centuries, was indeed characterized by what Fisher calls "quarantine." Although the empire was heavily involved in trans-Saharan trade, a trade that remained in Muslim hands, the rulers did not convert to Islam, and Muslims lived in a separate area apart from the neighborhood of the palace. On the other hand, the rulers of later medieval West African empires, notably of Mali and Songhay, but also of Borno and of the Hausa kingdoms, adopted Islam as the official religion of state. Finally, the late eighteenth and the nineteenth centuries saw a series of militant jihads launched by Muslims against these states or their successors, whose rulers were accused of being Muslims in name only, if at all. However, it must be pointed out that, seen in such chronological terms, these stages correspond, not to religious beliefs or practices, but rather to their role in the ideology of specific states. "Quarantine," "mixing," and "reform" are distinguished by the extent to which the legitimacy of rulers is couched in specifically Islamic terms. In Ghana, rulers made no claims to be Muslim. The adoption of Islam as the official religion of the court, in Mali for example, was intended to supplement rather than to replace other principles on which the rulers based their legitimacy. The jihads of the eighteenth and nineteenth centuries, on the other hand, were justified precisely by the identification of such

courts as fundamentally un-Islamic. However, whatever the ideology of rule, religious practice in any of these states and empires was essentially heterogeneous. Rather obviously, the characterization of a certain polity as in a stage of "quarantine" tells us strictly nothing about the nature of the beliefs and practices of the Muslim minority. Conversely, the leaders of jihads were generally unable to impose their views and practices on important segments of the population, notably much of the peasantry, not to mention dispossessed members of the former ruling class.

In any case, Fisher intends these stages to represent something more than a chronological sequence. The sequence is explicitly teleological; in Fisher's words, "the basic underlying progression has been towards a purer faith" (ibid.:31; emphasis mine). In the first place, such a Whig interpretation of Islamic history matter-of-factly takes an intrinsic tendency to progress for granted, without explaining why this should be so. (In a sense, once one takes such assumptions about progress for granted, only cases of backsliding need to be explained.) More disturbingly, such a teleological sequence implies not only that not all Muslims are equal, but that Western academics can determine which ones are more Muslim than others. In fact, notions akin to "purity" and "mixing" are categories of Muslim discourse, rather than objective categorizations of religious practice. Not infrequently, adherents of rival conceptions of Islam accuse one another of "mixing." In some circumstances, one party definitively gains the upper hand, as was the case in the successful jihads of the eighteenth and nineteenth centuries, where the adherents of one type of conception took over the apparatus of state power. To accept the "reformist" notions of "purity" and "mixing" at face value is to vindicate the jihadists, ignoring or dismissing the fact, for example, that large numbers of 'ulama' supported the other side. Any teleological approach implies that, in any fundamental debate among Muslims, one side represents a "purer" Islam than the other, and, what is more, that in the long run that side will win out. I must point out that nothing could be further from Fisher's intentions than such an asser-

tion. However, it is the only logical conclusion than can be drawn from an attempt to categorize different conceptions of Islam in terms of any developmental, much less teleological, sequence.

In all fairness, most recent scholarship on Islam in West Africa has avoided teleological assumptions that explicitly privilege one concept of Islam over another. To the extent than one can legitimately speak of bias at all, it emerges not so much in the work of individual scholars but in the choice of subjects that have received the most scholarly attention in the study of West African Islam. One of the principal foci of historical studies have been the eighteenth and nineteenth century jihad movements and the states they established. 'Uthman dan Fodio's jihad in what is now northern Nigeria has undoubtedly attracted the most attention, but jihad movements in Senegal as well as the jihads of 'Umar Tall and Ahmadu Lobbo in the Middle Niger have also been the subjects of considerable study.^[8] A second, and often related, area of focus has been the Sufi brotherhoods or orders: their diffusion to West Africa in the eighteenth and nineteenth centuries, their relationship to jihad movements, and their institutional forms, particularly where these have been highly structured and developed, as, for example, in the case of the Mouride brotherhood in Senegal, which has in and of itself received considerable scholarly attention.^[9]

Associated with these topical foci has been a regional focus on Senegal and on northern Nigeria, areas where jihad movements played a very prominent historical role and where Sufi brotherhoods continue to thrive and play a major role in local Islam. It would, of course, be churlish, and in any case unreasonable, to reproach individual scholars for choosing to devote their energies to such topics, which are as valid subjects of study in their own right as any others. Such studies have contributed greatly to the Western scholarly understanding of Islam in West Africa. I feel equally uncomfortable in suggesting that such studies might be the result of some sort of collective bias, as if there could be some conspiracy in the disciplines involved in the absence of conspirators.

However, it is easy to see why such subjects of study are attractive to scholars, for the jihad movements as well as the Sufi brotherhoods have, in the areas where they have flourished, made Islam very "visible" as a subject of study. The implications of this focus are most apparent in recent works surveying West African Islamic history as a whole (Clarke 1982, Hisket 1984), the bulk of whose discussions of West African Islamic history since the eighteenth century are devoted to jihad movements, to the spread of Sufi brotherhoods, and to the reactions of Muslims and Muslim societies to colonial rule (a subject described largely in terms of the continual spread of the Sufi orders and the rise of nonviolent, but militant, Muslim reform movements).^[10] A recent edited volume on The Cultivators of Islam (Willis 1979) highlights the same trends; the majority of the essays (if not, in certain respects, all of them) deal with the leaders either of jihad movements or of Sufi orders. Reading these works, one cannot help drawing the conclusion that reform movements of one sort or another on one hand, and highly organized Sufi orders on the other, have been the central features of Islam in West Africa since the beginning of the nineteenth century, and that anything else (except perhaps among recently converted "pagans," who cannot be expected to know better) is peripheral. I must stress, however, that such a bias, if it exists, has never to my knowledge been openly formulated. Indeed, I can perhaps justly be accused of reading it into works where it was never intended. In a sense, therefore, this criticism is distinctly unfair to authors who have attempted broad surveys. They have had to rely, for the most part, on secondary sources, and such sources are far more available about northern Nigeria and Senegal than about other parts of Muslim West Africa. However, one cannot simply assume that northern Nigeria and Senegal are in any sense "typical." More important, "typical" cases (whatever they may be) do not tell us any more (or any less) than atypical ones. The reform movements and the Sufi orders represent certain conceptions of Islam, but so do varieties of so-called "mixed" Islam where they exist and have existed, as well as yet other kinds of conceptions which may not fit neatly into any of these categories.

The fact is that such apparently "atypical" Muslim societies have not only existed in West Africa in the past, but that they continue to exist. The majority of members of these societies have been indifferent to, if they have not actively resisted, Muslim "reform" movements (though they have by no means been unaware of them). While certain members of these societies label themselves Qadiris or Tijanis, Sufi orders have never played a very central role in their conceptions of Islam. A whole belt of such Muslim societies, stretching from parts of Senegal to northern Ghana (including parts of Côte d'Ivoire, Mali, and Burkina Faso), are characterized by what Ivor Wilks (1984) has called the "Suwarian tradition," tracing its intellectual roots to the teachings of al-Hajj Salim Suware.^[11] (This is not to say that the Suwarian tradition constituted the only or even the major West African alternative to a stress on Islamic reform and/or highly structured Sufi orders; for example, Charles Stewart's [1973] study of the career of Shaykh Siddiyya al-Kabir in Mauritania does not focus on militant reform and consistently downplays the importance of Sufi orders, despite the fact that the shaykh was a Sufi leader.) If I insist on the existence of such alternate traditions in Islam in West Africa, it is not simply in order to demonstrate that the Muslims of Koko are not a "freak" of West African Islamic history, or that studies of such traditions are essential in the name of some criteria of exhaustiveness or of fairness. Rather, studies of Islam in such societies are theoretically important, not only in their own right, but also as a means of placing studies of Islamic reform and of Sufi orders in perspective. In the first place, a bias in favor of studying Muslim societies where reform movements have taken hold tends, as I have tried to argue, to give credence to the teleological notion (which does not follow) that such movements are historically inevitable, or at least that they represent the logical development of the process of Islamization (and by implication that Muslims who resist these movements are in some intrinsic way "out of step" with history). Secondly, and more important, there are epistemological limits to the kinds of generalizations one can make about phenomena such as reform movements simply by accumulating instances of cases where

they occur and take root. For example, one strategy for explaining such movements is to compare known instances and to attempt to determine the features they have in common. However, it may well be that some features that characterize societies where reform movements have been successful may equally characterize societies where such reforms have failed, or perhaps not even been attempted. To put it another way, one essential (but sometimes overlooked) component of the study of reform movements is to consider cases where reform movements have not met with success. Such cases constitute, if I may be permitted to use the experimental sciences as a metaphor, a kind of "control," a means of circumscribing the conditions under which Islamic reform movements will or will not take hold.

Choices and Commitments

In one way or another, both anthropologists and historians have had to cope with the problem of the multiplicity of conceptions of Islam. Very crudely, one might suggest that each discipline tackles a different dimension of the problem. History stresses the dimension of time, examining the succession of conceptions of Islam; anthropology, conversely, considers the question in space, contrasting different conceptions of Islam coexisting in the present, whether these be as remote from one another as Indonesia and Morocco or as proximate as different neighborhoods of a single city. Of course, this contrast between "history" and "anthropology" is caricatural, and as artificial as the division of fields of inquiry into the subject matters of various disciplines. In actual practice, as Ernest Gellner (1981: 214) has aptly pointed out, it is often hard to distinguish between the work of anthropologically minded historians and that of historically minded anthropologists. In a real sense, however, the problem remains the same whether one considers the dimension of space or of time: why do groups and individuals adhere to one conception of Islam rather than another? What constitutes a set of alternatives at any one time may emerge, in

hindsight, as part of a process of change, as groups abandon certain conceptions in favor of others.

One way of conceptualizing either the coexistence of alternatives or the process of change over time is in terms of the choices of individual actors. In a purely formal sense, religious commitments—like any other kinds of "decision"—are the outcome of individual choices. As long as individuals are aware of alternatives in the religious domain, whether this be between different (though not inevitably competing) religions or between different conceptions of a single religion, it follows that individuals have "chosen" the beliefs and practices to which they adhere. However, to frame the process in terms of choice is to imply that individuals engage self-consciously in making decisions about religious beliefs and practices. This is clearly the case in certain instances, and such decisions may involve a great deal of agonizing and soul-searching. Conversion, for instance, and indeed any deliberate act of change in religious belief and practice, can reasonably be labeled a "choice." On the other hand, it is one thing to be aware of, and indeed deeply concerned by, the existence of alternative religious beliefs and practices, and quite another to engage deliberately in weighing these alternatives in order to decide one's own commitments. For example, some of my Muslim friends in Koko would, from time to time, contrast their own Muslim practices to the practices of Christians as they understood (or misunderstood) them. What is more, they cited one—very isolated—instance of a fellow Muslim in a neighboring village who converted to Christianity (to evangelical Christianity no less!). In short, they were not only willing to consider the abstract possibility of Muslims converting to Christianity, but went so far as to acknowledge that this actually took place. Yet I am deeply convinced that the possibility that they themselves might convert to Christianity had never once occurred to my friends. Their arguments about the superiority of Islam over Christianity were not the outcome of any reasoned process of weighing alternatives, but rather a foregone conclusion from the very beginning. It would be thoroughly misleading to say

that they "chose" Islam in the same sense that the lone convert they mentioned "chose" Christianity.

Even when the outcome of individual commitments is not predictable from the outset, it does not necessarily follow that the actors are always conscious of making a deliberate act of choice, of weighing one alternative against others. It is often the case that individuals find certain arguments, certain points of view, more persuasive than others. Certain beliefs may make sense to certain individuals, while other beliefs appear to be nonsensical; certain practices are imbued with self-evident meaning, others appear as but empty gestures. It may not seem to make much difference whether individuals are persuaded by, rather than choose, one alternative or another. However, I would suggest that individuals both experience and verbalize "choice" differently from "persuasion," even though the outcomes may be the same. If choice is the outcome of rational decision-making—if individuals, in other words, behave like "economic men" even in the religious domain—then actors ought to be able to explain the costs and benefits of each alternative. On the other hand, it seems rather silly to ask individuals why they do not choose beliefs they find nonsensical or (unless they are in some way constrained to do so) they engage in practices they deem meaningless. This hardly implies any inability to verbalize the differences between various alternatives; quite the contrary. Such verbalizations, while they may certainly yield insights into the nature of people's commitments, are hard to take at face value, particularly if one tries to force them into a "rational choice" model.

The question remains: how can we best understand the real commitments of groups and individuals, particularly when, as we well know, change really does occur? Robin Horton's (1971, 1975a, 1975b) theory of African conversion probably remains the most systematic approach to this kind of question. Briefly, Horton argues that African cosmologies have a two-tiered structure, with a single supreme creator at one pole and lesser spirits at the other. (Horton never explains why this should pertain specifically to "African" cos-

mologies; there is no reason a priori why Horton's explanatory scheme should be more or less applicable in Africa than anywhere else in the world.) In particular, lesser spirits are associated with events within the social "microcosm," that is to say, the local community and its immediate social environment. The supreme being, on the other hand, governs the "macrocosm," and in particular relationships that transcend the immediate horizons of the community. To the extent that groups and individuals are critically involved in relationships beyond the "microcosm," they will consequently emphasize explanations of events that invoke the supreme being, rather than lesser spirits. In other words, polytheism and monotheism constitute poles along a single continuum, varying along with the degree to which social relations are internally or externally directed.

Clearly, Horton's scheme is far too mechanistic, but it is precisely the extent to which it focuses attention on social relations that makes it a useful point of departure. Unfortunately, Horton conceptualizes both social relations and types of religion in terms of single continua, from "microcosmic" to "macrocosmic" and from "polytheistic" to "monotheistic." It is by no means obvious that individuals and groups, much less "societies" as wholes, can be unambiguously or even meaningfully placed along a single "micro/macro" scale. The urban beggar and the bureaucrat both depend on the "macrocosm," but they live in very different worlds. Rating religions, or conceptions of religions, in terms of their degree of monotheism is an even more perilous enterprise in certain respects. After all, reform movements in both Christianity and Islam sometimes attack their opponents for harboring implicitly polytheistic beliefs or practices. According to Horton, a belief in the efficacy of saints, to take only one example, is relatively polytheistic in the emphasis it places on "lesser spirits" as opposed to the "supreme being." A great many Muslim and Christian theologians would take issue with, just as others would take comfort from, Horton's characterization. But is this a matter for secular academics to decide? In fact, Horton's scheme, like any other unilineal

scheme of development, is by implication teleological. Movement can occur in only one of two directions: forward or backward. While oscillation is clearly possible in the short and even the medium term, the long-term outcome is in no doubt.

What is more, such unilineal approaches are in the fullest sense reductionist, reducing everything to one single dimension. One feature of Horton's reductionism is that, for the purposes of his argument, he reduces religion to cosmology, and in particular to systems of "explanation, prediction and control." The central premise is both attractive and convincing: people will be attracted to a system of explanation that both makes sense of their environment and suggests ways of acting within it. However, such a reasonable assumption is only relevant to the study of religious change if cosmology, or at least system of explanation, is the major feature differentiating alternative religions or conceptions of religion. Horton's paradigm of conversion, adapted from John Peel's (1969) study of Aladura churches in Nigeria, revolves around the initial confrontation of mission Christianity and the traditional religions of converts or would-be converts. In many respects, this represents an extreme case. Early missionaries were, by and large, totally contemptuous of African cosmologies, generally dismissed as "heathen superstitions," and deliberately set out to replace local beliefs as well as practices.^[12] As long as missionaries insisted on the strict incompatibility of African and Christian cosmologies, Africans had no choice but to decide between one or the other, or else—as was ultimately the case with the Aladura—to devise their own alternatives.

However, cosmology was hardly the only factor differentiating Christianity from traditional African religions, and it is by no means obvious that it was the deciding factor in accounting for the preferences of all the groups or individuals who converted. Often enough, cosmological concerns are quite secondary. When I was first in Koko in 1972, a religious controversy was raging within the Muslim community. The

most heated discussions revolved around whether or not one should cross arms in prayer. All told, Islamic controversy in Koko tended to concern ritual—which rituals were proper or improper for Muslims to perform, and how. Such questions of ritual, much more than of cosmology, distinguished (and continue to distinguish) the Muslims of Koko from their "pagan" neighbors. Nor is it possible to argue that ritual is essentially a mirror of cosmology; the position in which one prays is certainly not, in any straightforward way, related to any particular explanation of the world. It would be much more to the point to argue that arguments about ritual are also arguments about morality—that is to say, about how one should or should not act in specific situations.

Indeed, issues of ritual and morality are just as salient as cosmology in discussing the differences between mission Christianity at the turn of the century and traditional African religions. Horton's self-styled "intellectualist" approach assumes that ordinary people behave rather like Western academics (or perhaps as Western academics would like to believe that they themselves behave), and that their primary concerns are with finding the best possible way in which to explain the world. Seen in this light, controversies about how to pray are either entirely irrational or else simply smokescreens for some more "fundamental" debate about explanatory principles. Intellectual concerns are simply one of several dimensions that allow one to contrast different religions, or different conceptions of the same religion, and it cannot be assumed that they are more important than the spiritual, aesthetic, ritual, or moral dimensions of religion. Rather than prejudge the question, as Horton does, it is essential to determine, in the first instance, the specific nature of the controversies that oppose one viewpoint to another. If the groups and individuals involved insist that ritual is of crucial importance, we must assume that this is so and accept this assertion at face value. What still needs to be explained, of course, is why arguments arise over specific issues, and why groups and individuals commit themselves to one side or the other.

I would suggest that one can be even more directly sociological than Horton without necessarily being nearly as reductionist. Religious conceptions are not only conceptions of the world but conceptions of society, or, to phrase it in a less Durkheimian manner, of how individuals ought to behave toward one another, as well as toward God and/or toward spirits. These moral and social dimensions of religion cannot be reduced to abstract moral rules such as "Thou shalt not steal," and "Love thy neighbor as thyself," let alone to the prescription that one pray five times a day in the direction of Mecca, though such rules are obviously important, too. Rather, questions about how individuals should behave toward one another frequently require that one ask the subsidiary questions "Who?" and "Toward whom?" In other words, rules of behavior are formulated in terms of social identities: "senior" and "junior," "male" and "female," "insider" and "outsider," and so on. Two truisms need to be stressed: first, that all individuals have multiple social identities; second, that specific social identities are only salient in contrast to other possible identities, although these do not necessarily (as the very partial list above might imply) have to come in pairs. I would like to suggest that the choices individuals make between different conceptions of religion, or for that matter between different religions, have a great deal to do with the relationship between religion and social identities.

In the first place, religions (at least monotheistic ones) entail the adoption of specific social identities. For example, to be a "Muslim" means, among other things, not to be an "unbeliever," though whether an unbeliever is Christian or "pagan" also makes a difference. Different conceptions of Islam may draw the distinction between "Muslim" and "unbeliever" in different ways. Indeed, during the jihads of the eighteenth and nineteenth centuries, it was not unusual for the battle lines to be drawn according to the ways in which "Muslim" and "unbeliever" were to be defined. In any case, "Muslim" and "unbeliever" are not the only social identities defined in obvious ways by conceptions of Islam. Other iden-

tities may distinguish a specific subset of Muslims from ordinary believers, and in some cases from one another; "scholars," "sufis," "saints," and "sharifs" are examples. The importance accorded to any such category, the kinds of behavior expected from individuals who belong to them, the ways in which individuals may lay claim to belong to them, and the way in which such claims are likely to be accepted or rejected by their fellow Muslims may all differ from one conception of Islam to another. In short, one can reasonably ask questions like: What does it mean to be a scholar or a sufi in Koko? To what extent has this meaning changed over the past fifty years? To what extent is this meaning challenged in one way or another by alternate viewpoints within, or at least known to members of, the community?

In important respects, such identities are specifically Islamic, even if they are not common to all conceptions of Islam and if analogues can be found within other religious traditions. They can all be labeled, in some commonsense if not theoretical way, as "religious" identities, and any reasonably complete description of Islam in any particular community or society, inside or outside of West Africa, can at the very least be expected to record their presence, if not their absence. However, specifically "religious" identities obviously do not exhaust the realm of relevant social identities in any society. Individuals also have specific ethnic, political, and kinship affiliations. They are male or female, slave or free, elder or junior, warriors, traders, or farmers, and so forth. The list of such relevant identities obviously varies from place to place, time to time, and even context to context. More important, not only the nature of such identities, but the ways in which they are defined, the implications of claiming one identity (or being ascribed it) rather than another, may also differ. Many of the discussions of subjects such as ethnicity, class, and gender in modern social science hold these propositions about identities to be self-evident. However, such identities are also directly relevant to the study of religions such as Islam. Conceptions of religion do not necessarily reflect, in any mechanical way, "nonreligious" social identities, but rather

have in one way or another to come to terms with them, as features of the social universe that cannot simply be ignored. An obvious example is the relationship between different conceptions of Islam and notions of political legitimacy. At one extreme, in a theocracy, a leader's legitimacy may be defined in exclusively religious terms. The modern Iranian regime is an example.^[13] More commonly, conceptions of Islam may legitimize a ruler's authority, even though this authority may not be defined exclusively in religious terms. The West African "theocracies" established by various jihad movements are examples; succession to authority was determined, in these polities, by various hereditary principles. In a very different vein, other conceptions of Islam uphold the legitimacy of the Sharifian dynasty of Morocco. Even the dynasty's most fervent supporters would not argue that descendants of Ali are the only legitimate rulers of Islamic states, but only that, all other things being equal, they are especially qualified to rule. On the other hand, certain conceptions of Islam may simply tolerate political authority, perhaps as a necessary evil. Even here, there is room for important differences, depending on whether or not it is important or indeed necessary for the ruler to be a Muslim (and, if so, on how one decides whether or not he is in fact a Muslim). Finally, of course, conceptions of Islam may deny the legitimacy of particular rulers, though again it matters whether this denial ought to be translated into a stance of quietism, of passive resistance, or of active revolt. One way or another, every conception of Islam must adopt one such stance. Seen in this light, professed indifference to principles of political legitimacy is not the absence of a position, but simply one possible position among others.

The issue of political legitimacy comes, of course, as no surprise to students of Muslim societies. The study of Islamic theories of the state is a reputable and firmly established scholarly subspecialty. However, I wish to suggest that ethnic differences, class differences, age differences, gender differences, and the like are all equally relevant to different con-

ceptions of Islam wherever such differences are socially salient. Different conceptions of Islam must either in some ways legitimize these distinctions, refuse to acknowledge that such distinctions can legitimately be made by Muslims, or pass over these distinctions in silence on the grounds that they are ultimately irrelevant to Islam. Different conceptions of Islam are consequently not simply different explanations of the universe per se, but rather different ideologies of the social universe. By calling such conceptions "ideologies," I merely wish to underscore the point that they either confer or deny legitimacy to specific social distinctions. Even silence about such distinctions confers a sort of provisional legitimacy, for such silence implies that any present state of affairs is at least tolerable as far as Islam is concerned, although at another level it may also free Muslims to contest such distinctions without necessarily running the risk of violating religious principles.

By suggesting that conceptions of religion are, in this broad sense, ideological, I certainly do not wish to advocate a narrowly instrumentalist approach to understanding religion, either by implying that individuals deliberately choose a religion or conception of religion that is in their own best interest or, even more cynically, by suggesting that religion fosters the interests of the powerful by gulling the powerless into accepting their inferiority. Such approaches to religion are modeled on one variety or another of economic theory, classical or Marxist as the case may be. The classical model does not explain how individuals actually decide what is in their own best interest; the Marxist model leaves us with the mystery of why the oppressed are so often and so easily hoodwinked by their exploiters.^[14] This is hardly to deny that there is any relationship between religion and the "interests" of individuals in society (though these interests are more usually identified by academics than by the individuals concerned). However, the relationship is a more complex one than simple instrumentalist explanations would lead us to believe.

Rather, I hold that different conceptions of religion define and express the ideal nature of communities and consequently the proper place of individuals within them. Hierarchies of one sort or another are legitimated or contested, the responsibilities (or lack thereof, in some instances) of different categories of individuals toward one another are delineated, principles of authority of various kinds are established. Religions provide one, though by no means the only, category of answers to questions about who individuals are, who they ought to be, and whom they can aspire to become with respect to one another. Religion does not "reflect" society but rather makes sense of it, morally as much as intellectually.

Precisely because moral issues, in the most general sense, are so central to the differences between conceptions of religions, a model of choice based on a deliberating rational actor is inappropriate. Only cynics are conscious of "choosing" in such matters. Individuals adhere to one position or another because they are convinced that it is "right." Such "rightness" entails first of all that an ideal of a community or communities be salient within the social environment of the groups and individuals concerned. For example, an emphasis on genealogy may simply be irrelevant in a heterogeneous and relatively anonymous urban setting; it makes little sense to define oneself with reference to one's ancestors if no one else knows who they are. On the other hand, "rightness" is also related to an individual's sense of his or her own worth. Individuals will more likely be attracted to a vision or a moral universe that vindicates them for being who they are rather than relegating them to a relatively marginal moral existence.

It follows from such an approach that religious controversies revolve in fundamental ways around different ideals of "community" as they apply to concrete communities of real groups and individuals. Different social categories of individuals, then, are more likely to be attracted to one point of view or another. The stakes, in a very real sense, are the communities in which they live, for these communities are not simply "given," but rather are socially constructed entities

predicated precisely on the collective recognition of some common moral framework. This common moral framework is always subject to renegotiation; it can always be called into question in one way or another. Such processes are virtually inevitable; communities do not exist in isolation but are parts of wider social environments subject in their own right to processes of short- and long-term historical change. The vision of "community" at stake in religious controversy is not simply the web of relationships between its members, but equally the nature of the relationship of the community as a whole, as well as of its individual members, with different institutions, groups, and communities in the outside world.

This is broadly the perspective from which I shall analyze religious controversy and change within the confines of a single community over roughly the past hundred, and especially the past fifty, years. The key question informing my analysis is, quite simply, what is at stake? In the first place, the question needs to be taken absolutely literally: what, specifically, are the issues of controversy or of disagreement within the Muslim community? For example, does one pray with arms crossed or outstretched? The matters over which people disagree are also those they consider to be the most important. It is not up to academics to decide which issues are trivial and which weighty. However, by asking what is at stake, we need to understand not only what the source of disagreement is, but ultimately "To whom?" and "Why?" The way I address such issues necessarily reveals my own methodological and theoretical biases. In the first place, as I have argued above, it is misleading to understand individual commitments as the outcomes of processes of deliberate choice—that is to say, as formally equivalent to decisions about which breakfast cereal to purchase. This is certainly not to imply that such decisions are irrational; they are neither more nor less rational, I suspect, than most other decisions humans have to make in the course of their lives. However, I would argue that the justifications that people give for their commitments to one side or the other are not necessarily identical to their underlying reasons, reasons they may not necessarily

articulate either to anthropologists or to themselves. Justifications are context-bound; they depend on who is providing them, and to whom. Clerics are likely to give different kinds of answers than ordinary believers, and indeed to frame an answer differently if addressing a colleague rather than a lay Muslim seeking counsel, or an adherent of an opposing point of view. Gender, the relative seniority of the speakers, the presence or absence of an audience, all of these are among the factors that may influence the justifications any individual may or may not provide in a given situation. Needless to say, answers to the prodding questions of anthropologists or other outside observers are even more problematic. This is not to say, of course, that one should ignore the justifications individuals provide for their own religious commitments, but rather that such justifications may raise as many questions as they answer.

In the last resort, I have looked for underlying patterns in terms of the categories of persons—in terms, for instance, of age, class, gender, ethnicity, and level and nature of education—who tend to commit themselves to one side or the other of any controversy. There is no preestablished list of relevant categories, nor is there any a priori way of decoding in any particular case which categories will be salient. This is precisely because the controversies revolve around the religious relevance or irrelevance of certain social distinctions, around the legitimacy or illegitimacy of different hierarchies and of different forms of authority. Different conceptions of Islam, by the ways in which they confer or deny legitimacy to social relationships, provide individuals with different possible ways of defining their own identities and the identities of others. It is presumptuous for an outside observer to decree which of these ways is ultimately in anyone's best interest, and virtually impossible to determine whether, much less how, individuals have attempted any such calculations themselves. The best one can do, perhaps, is to try to show why they might find either one side or the other more persuasive, more in keeping with a vision of the social universe where they have more than a marginal place.

Admittedly, this view of religion is a variety of sociological reductionism. In focusing explicitly on the social dimension of Islam, I certainly do not wish to deny the importance of its other dimensions—intellectual, spiritual, and aesthetic. The commitments of specific individuals may be informed by any or all of these concerns, rather than by the nature of the legitimacy one conception or another of Islam confers on specific visions of "community." However, my analytical concerns are ultimately not with the religious commitments of individuals per se, but with those of a community as a whole. Certain very real changes have taken place, changes of which members of the community are fully conscious, and with which they are able to identify. Others have equally really been attempted and have failed to win support, and these failures are perhaps as significant as the successes, for they show that while changes of one sort or another are inevitable, it is impossible to predict with certainty of what sort they will be, and to identify a single "direction" in which Islam is heading. Any teleological scheme accounting for change—and all directional schemes involving some notion of development as an explanatory principle are teleological—is ultimately reductionist, whether or not it is sociological. If real historical changes in specific communities at specific times are not simply the expression of intrinsic and inevitable forces pointing in a single direction, neither are they simply the aggregate of individual "decisions" made in isolation. These are issues of concern to everyone in the community. The actions of individuals tend to generate opposite, though not always equal, reactions. The adoption or rejection of changes by groups and categories of persons, whether they represent an abrupt switch from one conception of Islam to another or rather a more gradual shift in the orientation of any specific conception, are "social facts," and I would argue, with Durkheim, that as such they are to be understood in social terms.