Version classiqueVersion mobile

Comparer en histoire des religions antiques

Claude Calame
Bruce Lincoln

Comparison and the Study of Religions of Late Antiquity1

David Frankfurter


Dans l’étude du monde méditerranéen dans l’Antiquité tardive la comparaison s’impose en tant que moyen pour justifier des généralisations relatives à la nature de la religion, au chistianisme, et aux traditions religieuses devenues caduques ; c’est aussi un moyen pour développer des contextes idéaux ou des modèles dynamiques pour rendre compte des anomalies dans la documentation historique ou archéologique. À partir de l’illusion de l’analyse du matériel historique en termes purement « émiques », l’étude passera en revue les méthodes de comparaison explicite et implicite proposées par V. Bonnell, J.Z. Smith, and P. Brown, puis par l’auteur lui-même, dans différents cas de figure. L’usage de « patterns » pour rendre compte d’anomalies historiques et pour développer des comparaisons transculturelles, loin de déformer la documentation première, offre des contextes théoriques pour mettre à l’épreuve des cas historiques.

Texte intégral

  • 1 I am immensely grateful to Claude Calame and to the Casablanca Group for comments in the revision o (...)

1This essay seeks to introduce and justify methods of comparison in the study of religions of late antiquity, a field distinctive for including — or for many scholars, framing — the rise of Christianity. A comparative approach is explained in which primary materials are understood first in the context of systems and then as examples of some larger pattern. That pattern provides the means of generalizing across two or more further examples from the same historical period, but it might be more richly exemplified in some geographically or historically removed (even modern) examples. By reference to such larger patterns or categories, obscure or challenging phenomena can be understood in familiar terms, ostensibly familiar phenomena can be understood in richer and often more challenging contexts, and — most importantly — historical interpretation itself, which relies intrinsically on the phrasing of indigenous (“emic”) ideas and actions in modern critical (“etic”) categories can operate at a more responsible and constructive level. The third and fourth parts of this essay will explore ways in which comparison has aided work on the religions of late antiquity.

Challenges and Objections to Comparison in the Study of Late Antique Religions

2The documentation for religion in late antiquity challenges us in multiple ways : the invention of broad discourses of empire and salvation ; the nature and dynamics of Christianization ; and the beginnings of indigenous notions of “religion” and anti-religion or heresy. But it is also an area of study that has been, perhaps more than any other, distorted by triumphalist, romantic, and theologically-motivated scholarship, both in the uncritical nomenclature of “church” and “paganism” to which most historians still cling, and in the assumptions about conversion, decline, and religion itself that still undergird most late antique studies.

  • 2 See, e.g., Luck, 2000, p. 223 – 38 ; Cameron, 2011, p. 3. See also Ascough, 2009.

3Part of the blame for the uncritical perpetuation of theological bias lies in the very status of the period as one of Christian triumph, pagan decline, or the “paganization” of apostolic Christianity — in short, a transition between two morally-loaded states of culture. This was the way Edward Gibbon saw it in his time, William Ramsay and Alphonse Barb in theirs, and in different ways, a number of scholars today.2 Even as blatantly theological motivations have subsided in most work on religions of the Roman empire, the same terms and spiritual criteria have so thoroughly influenced the field that we continue to read them in discussions of “paganism,” “mysteries,” and “syncretism.”

4The context in which so much scholarship on religion in late antiquity maintains these theological undertones can be found in the very self-conception of fields like classics, history, and archaeology, out of whose ranks most late antiquity scholars come. These are fields that have traditionally imagined themselves as empirical, dedicated only to primary documents and voices that they merely interpret. The notion that interpretation and historical inference, especially about broader themes like religion or Christianization, might be informed by assumptions tends to be discounted.

  • 3 See, e.g., Lane Fox, 1987 ; MacMullen, 1997.
  • 4 Echoing Cameron, 2007, p. 43.

5Except for synchronic contrasts (or elisions) of cultures or Christianities within the late antique world in the service of historical generalization, comparison has been altogether unusual as a method in the interpretation of this period.3 One might call this a strange, even ludicrous avoidance, since other pre-modern cultures also integrated Christianity and assimilated Christian myth and ideology in popular, elite, and economic domains. If our historical interest lies in religious transformation and Christianization, then how could the late antique Mediterranean and Europe conceivably be unique ? Mayan cultures, Andean cultures, and African cultures all provide exceedingly well-documented cases and sophisticated ethnographic investigations of Christianization in action. The dynamics of Islamization and Buddhization at the regional level have much to teach the student of late antique Christianization in the areas of thaumaturgy and charisma, textuality and magic, and cults of images and relics. But historians of late antiquity have long shied away from investigating how similar processes might have gone on elsewhere, invariably out of fear of dilletantism, linguistic inability, even anxiety about what to do with ethnographic comparanda. What could a “similar case” in Guatemala possibly prove ? Couldn’t you find a comparandum to “prove” anything you want?4 Who would want to be identified with the excesses of the comparativists of yore, for whom every likeness revealed some theological notion ? And isn’t there enough work to do in one’s historical field without reading about everything else ? How much time does anyone have in the day ?

6Obviously these responses misunderstand the reasons for comparison — indeed, the inevitability of comparison. For comparison is the very foundation of generalization, which historians do habitually, with every second-order term they use or system they observe. Our use of “religion,” “sacrifice,” “magic,” “amulet,” “canon,” “ritual,” or “shrine” is not a simple “emic” translation of some unambiguous Greek or Latin or Hebrew word but a second-order, heuristic category of classification that implies applicability to a particular spectrum of like data. And to the degree that we historians of religion tend to latch onto historical situations in which “canon” or “votive deposit” or “sacrifice” assume some hybrid form or instigate some historical controversy, we do well to read widely in the dynamics of such a category — such a phenomenon in religion — to get accustomed to its parameters and tendencies cross-culturally.

The Method and Inevitability of Comparison

V. Bonnell : Analytic and Illustrative Comparison

7As Victoria Bonnell laid it out in her classic 1980 article, there are two types of useful comparison. In the first, which she labels the analytic mode, one examines several different historical incidences, usually linked by major cultural variables, to identify specific differences. We might think of the different local responses to Satanic sorcery in sixteenth-century Germany, Poland, Sweden, Switzerland, and England, or the differing functions of church liturgy between seventh-century Byzantium and the western Catholic Church. No broader interest is expressed in, say, the general parameters of constructing and expelling evil in local cultures, on the one hand, or the general functions of ritual space and architecture, on the other. The analytic comparison is relatively bounded in space and time and usually serves the historical reconstruction of a particular period : e.g. during the seventh century, while eastern basilicas sought to cultivate a sense of the heavenly world with icons and incense, Latin churches became the locus for reproducing Christ’s sacrifice in the Eucharist. The categories used are often specific to the period : liturgy, Eucharist, Sabbat, icon. There is generalization, of course, but the generalizations are discrete to the period or the religious formation — to the identification of independent variables — and no claims are made to contributing to the understanding of religion itself.

8The other mode of comparison, which Bonnell calls illustrative, seeks to frame, articulate, and rectify a general pattern that can be used (and further rectified) to highlight historical anomalies in further cases and, ultimately, to gain a better sense of the meaning of the pattern itself. The individual cases serve ultimately as examples of the pattern, and so there must always be more than two cases. A pattern can only be extrapolated from more than two cases, although the historian might only want to test it in one particular case.

J.Z. Smith : Examples and the “Third Term” of Comparison

  • 5 See Smith, 1982, p. xii – xiii ; 2004, p. 230 – 250. Examples : Smith, 1978, p. 147 – 171 ; 1982, p (...)

9This mode of comparison has been principally associated in recent decades with the work of Jonathan Z. Smith, the American historian of ancient Mediterranean religions who, notably, prefaced his academic career with extensive work in botany. In a series of essays and published lectures over the course of the 1970s and 1980s Smith developed a set of principles for meaningful and productive comparison of religious phenomena that sought to move comparison beyond the theologically-motivated efforts of prior scholars, especially those who focused on Christianity, Judaism, and so-called “paganism” in the ancient world.. The foundation of these principles was the historian or interpreter’s readiness (a) to comprehend the anomaly, the Other, and (b), by a series of comparative steps, to render that anomaly familiar (Smith’s own examples included the inverted crucifixion of Peter in the Acts of Peter and the mass-suicide at Jonestown, Guyana, in 1978). By the same token, he insisted, the familiar datum — the Gospel text, the votive stela, or the incised cross — must be itself constructed as anomalous and then re-familiarized through comparison if it is to be understood as anything but a theological projection.5

  • 6 See Smith, 1982, chap. 2.
  • 7 See esp. Smith, 1990, p. 99.
  • 8 Referring to the critique by Sandmel, 1962. See Smith, 1990, p. 51 – 53 ; 2000 ; and Paden, 2000.
  • 9 Smith, 1990, p. 52 ; see also, 2004, chap. 8, esp. p. 194.

10How, then, to proceed comparatively ? First, Smith argued for the comparison of systems, not isolated and acontextual words or images : thus, for example, meal rituals rather than bread ; apotropaic image rather than the cross.6 Then, in a way paralleling Bonnell, he insisted that comparison always take place in relationship to a “third term” — a taxon or pattern — which prevents the implication (or subsequent proposition) of borrowing or influence. This principle, he argues, is especially important for the study of Christianity among the religions of the Mediterranean world, since scholars have typically assumed that parallels between Christian ideas and (say) Mithraic or Dionysiac materials implied Christianity’s dependence on these religions, no matter the relative historical context of the data.7 Finally, comparison among data should always proceed in relationship to that third term, and preferably with more than two comparanda. It means little to show that x resembles y ; it means a lot to demonstrate that x resembles y more than z with respect to [pattern X], for this strategy admits the multiplicity of potentially comparable data at the same time as the important differences (context, history, social world) that prevent mere “parallelomania” — that is, the gratuitous juxtaposition of similar data without regard for context, difference, or the meaning of similarity.8 For example, the cult banquet for the Sarapis súnodos in Greco-Roman Oxyrhynchos may indeed resemble certain early texts about the Christian Eucharist more closely than the stylized Jewish ritual meals at Qumran with respect to the sacred meal in antiquity — or more precisely the symbolizing of meal components in the course of the sacred meal. We begin to learn about the variety and taxonomy of “sacred meal” through the relative differences among these exempla. “That is to say,” Smith points out, “comparison does not necessarily tell us how things ‘are’… ; like models and metaphors, comparison tell us how things might be conceived, how they might be ‘redescribed’… Comparison provides the means by which we ‘re-vision’ phenomena as our data in order to solve our theoretical problems.”9

The Generalized Pattern as Interpretive Goal

11Comparison in this sense is ultimately conducted between proposed pattern (meal rituals, magical gesture, public ceremony) and situation or datum (Book of Revelation, Qumran document, votive inscription, historical narrative) rather than between one datum and another. Furthermore, the pattern — even the very framing of the datum under discussion — is constantly adjusted in order to determine relevance : should we be talking about “magical gestures” or rather apotropaic gestures ? What about when early Christian people cross themselves : protective, purifying, or a gesture of identity ? Should we be sub-dividing the concept of “sacred meal” in order to make sense of some major variations ? By what general terms should we describe the angelic songs, bowls, and trumpets in the Book of Revelation so that we can allow for the author’s gesture to both Jewish temple priesthood and the public ruler cults of Asia Minor ?

12The comparative enterprise is thus far more than simplistic juxtapositions but covers the entire dialectical process by which we define what it is we are talking about with religious phenomena, from the translation of simple “emic” terms (mágos, euanggélion, eulógion) to the operable generalization about actions, attitudes, structures, or religious processes (“sacrifice,” “prayer,” “syncretism,” “shrine”). Each term depends on a cluster of exempla, some of which might be regarded as typical and some as anomalous, and that term thereby dictates how further data will be understood, labelled, and brought into the cluster. We may restrict the cluster to a single archaeological site, a single historical period, a single region over several centuries, or — as is typical with studies in religions of late antiquity — the entire Mediterranean world from the first through fourth centuries. We may also open up the cluster to modern ethnographic examples that might, as I shall explain below, offer richer, “living” contexts for religious phenomena than ancient archaeological or literary materials by themselves can afford.

13There is a humanistic aspect to this type of comparison. That is to say, one’s eyes are on the broader intellectual understanding of phenomena, not simply the particulars. As the entomologist studying a particular beetle in New Guinea is ultimately concerned with the articulation and rectification of Coleoptera, the order of beetles, so the pattern-comparativist may ultimately be interested in the patterns as illustrative of human cultural dynamics — religious, economic, political, even psychological. We engage in it for the greater understanding of human society. For this kind of illustrative comparison our own specialty areas — through which we investigate patterns comparatively — really constitute a kind of ethnographic fieldwork for the larger understanding of religion.

14We do not, of course, delude ourselves with the impression that the patterns exist apart from their heuristic function in making sense of religion in context or that they grasp in any way the totality of content or experience. They simply aid us in making sense of phenomena and in bringing our observations to new situations.

Comparison and the Fallacy of “Emic Interpretation”

  • 10 See esp. Smith, 1990.
  • 11 On the category “sacrifice” see now Frankfurter, 2011.

15While Bonnell and Smith both associate comparison with this broader mandate, the observation, definition, and rectification of patterns, it is extremely important (for historians and classicists in particular) to understand the influence of comparison — whether critical or implicit — on the very terms we use to translate key words or interpret situations. We may tend to assume that in discussing mageía or sacerdos or thusía or sacrificium as “magic” or “priest” or “sacrifice” we are reflecting the indigenous, “emic” sensibilities of the Greeks or Romans whose voices we are interpreting, but this is quite erroneous. To translate is inevitably to lift a term from its “emic” sense into an often heavily loaded, modern, “etic” category. “Magic,” “priest,” and “sacrifice,” for example, are irrevocably tainted through implicit comparison to Christian tradition — indeed, Protestant critiques of Catholic tradition10 — and cannot “simply” translate Greek, Latin, or other primary terms for some ambiguous area of ritual power, some acquired ceremonial leadership role, or a range of offering traditions that included ritual animal slaughter.11

  • 12 Many of the criticisms Claude Calame marshals against anthropological interpretation (2002) are all (...)

16That is to say, we deceive ourselves when we imagine ourselves to be working on historical or textual materials purely in indigenous terms — “as the Greeks/Romans/Egyptians understood this matter” — for our translations and interpretive discussions lift indigenous terms and perspectives out of their world and into a modern context in which they can only gain meaning and value for discussion through comparison. And if that comparison is not constructed with critical attention to the general and specific terminology by which an ancient phenomenon is rendered intelligible, then it will be constructed by default through unconscious comparisons that lie close at hand in the culture : terminology drawn from Protestant or Catholic Christian tradition, exotic stereotypes of priests or “pagans,” and for “sacrifice” romantic notions of Christ and selfless mothers. If criticism provides the means by which scholars consider the biases and implications of their categories and terms, comparison (in the sense developed here) allows the rectification and consistency of those categories and terms, such that a study of the “priesthoods” in Roman Oxyrhychus, the uses of ritual animal slaughter in Hellenistic Phrygia, or the use of binding spells in fourth-century Rome actually furthers thinking in the field of ancient religions.12

Conceptualizing Religion in Late Antiquity Through Comparative Methods

The Potential of the Ethnographic Comparandum to Rectify Historical Themes

  • 13 See, e.g., Dodds, 1965 ; Lane Fox, 1987 ; Stroumsa, 2009.

17Great historical writing always reveals larger themes and underlying trends, whether by reinterpreting well-known data or introducing new data. The question then becomes how these themes are developed, justified, and articulated, since the “emic” terminology of the historical insiders is never sufficient to substantiate a theme that covers many regions, cultures, or periods. Beyond such critical questions as the process and progress of Christianization, the transformation or diminishment of traditional cults, and the political and economic aspects of both, historians of late antique religions have long proposed such general trends — religious patterns, in fact — as a rise in cultural anxiety or superstition, transformations in sacrifice and scripture, new constructions of holy men, asceticism and new constructions of sexuality, a desperate syncretism with multiple religious traditions or, conversely, new types of religious totalism and intolerance, such as iconoclastic violence.13 Each of these trends involves a religious pattern of sorts, generated out of impressions of, and then selections from, the data for religion in late antiquity. These historical patterns are therefore, ipso facto, comparative productions and therefore subject to the same testing by reference to further comparanda as any comparative pattern : was asceticism really new ? How does one describe the new attitudes to the body ? How does a Christian saint differ from an itinerant Jewish prophet ? Is “anxiety” or “superstition” a viable “etic” category for the description of late antique religious phenomena ? How representative is the voluminous epigraphical and iconographic evidence for theological syncretism in the Roman world, and what kind of pattern does it really reveal ?

  • 14 See Smith, 1978, p. 172 – 189 ; Frankfurter, 1998, p. 198 – 237 ; 2000a.
  • 15 See, e.g., Frankfurter, 1998, p. 131 – 142.

18These larger questions of description, terminology, and generalization demand a second level of comparison, in which examples offering richer, more dynamic context, even if more historically removed, are sought out to test the patterns developed for late antique materials. New developments in the self-representation of Egyptian temple priesthoods might, for example, be clarified through comparison to traditional ritual experts of other cultures forced to reinvent themselves to maintain authority in a new cultural situation.14 Such a comparison would sharpen our attention to aspects of the ancient historical data we might otherwise have missed, our sense of the dynamics surrounding the ancient historical data (perhaps limited to inscriptions or fictional narratives), and the language we use to describe the process. The activities of the domestic cult and its relationship to civic cult (or pilgrimage shrine) in Roman or late antique Ostia, Panopolis, or Sepphoris might likewise be clarified and enriched by comparison to living traditions of family altars and their cycle of decoration.15

Peter Brown : Social Anthropology and the Articulation of Religious Patterns

  • 16 Brown, 1970, following Douglas, 1966, p. 94 – 113. See also Brown, 1978, chap. 2, and Brown, 1998, (...)

19This approach to the generalizing of religious patterns in late antiquity is especially represented in the work of Peter Brown. Brown’s work on the social and religious world of late antiquity has been so inspiring to several generations of scholars for its unusual sensitivity to voices and realities once deemed marginal or decadent : the holy man, the cult of the saints, the strange abhorrence of the body in Christian asceticism. In these respects Brown is typically regarded as an historian of the late Roman and medieval periods. But many of Brown’s distinctive insights have actually involved the application of Durkheimian social theory — via Mary Douglas’s work — to the data of late antique Christianity through a series of implicit comparisons, in which his formulation of patterns in the religious transformation of the late antique Mediterranean world reflected social theories of charisma, the body, and social structure developed in the world of anthropology. Indeed, we can track Brown’s application and development of Mary Douglas’s anthropology from his 1970 essay on the location of malevolent powers in the early Christian empire, which drew closely on Douglas’s essay “Powers and Dangers,” to his 1988 book The Body and Society, which followed the insights of her 1985 Natural Symbols.16 Brown uses Douglas’s formulations of social theory to sensitize himself to data he might otherwise overlook, to develop a descriptive language for patterns he observes in the late antique data, and to articulate the patterns themselves as important and real features of religious experience in late antiquity.

  • 17 E.g. Brown, 1982, p. 118, 123, 124, 131.
  • 18 Brown, 1978, chap. 3.

20Perhaps most indicative of this method is his 1971 essay, “The Rise and Function of the Holy Man in Late Antiquity” (1971/1982), in which Brown articulates the social and economic roles of Christian holy men in their immediate milieux according to patterns generated by reading in North African and Siberian ethnography. That is to say, by reference to modern anthropological work that stressed the dynamics of exorcism, of emotional dramaturgy, of small community tensions, and of the authority of the peripheral prophet in negotiating a diversity of social tensions, Brown laid out how holy men in late antiquity likewise functioned in their social environments as catalysts of religious change.17 They were (as he put it in a subsequent book) “friends of God,” setting the Christian phenomenon in terms of — and according to the observed social dynamics surrounding — north African Muslim saints, who are known as awliyā, “friends [of God].”18

  • 19 Brown, 1998, p. 303.
  • 20 See, e.g., Kaplan, 1984 ; Stirrat, 1992 ; Howard-Johnston & Hayward, 1999 ; Frankfurter, 2003.

21This mode of comparison involves very much the Bonnell/Smith, “illustrative” comparative model described earlier, insofar as various ethnographic sources are read as reflecting a general phenomenon, or taxon, of charisma in society and its varying dynamics, for which the Christian holy men of the late antique eastern Mediterranean also serve as examples and for which all the examples offer critical rectification of a category “holy man” in particular historical situations. Brown later reflected that his late antique materials were not significantly different from the “tribes” — the diverse peoples in ethnographic reports — that Douglas herself used so adroitly as exempla to develop social theories.19 To be sure, Brown generally couched his comparisons, indeed, his very use of ethnographic comparanda, in the background of his published work (while recommending to his students a vast range of comparative materials to supplement and frame their historical interests). And it is thus testimony to the historical utility and accuracy of the models he generated through comparison that his approaches to the holy man, cults of saints, and the social symbolism of the body (among other patterns) have become the foundations of subsequent work on Christianity in late antiquity, continually applied, tested, and adjusted with new materials and new regions, and have even been extended to phenomena in medieval Ethiopia and modern Sri Lanka.20

Comparison by Reference to a Social Model

  • 21 E.g. Gager, 1982 ; Meeks, 1983 ; Kraemer, 1992.

22In his use of Mary Douglas’s comparative social theory, Peter Brown exemplifies the value of the social model itself in this kind of second-order comparison. That is to say, the model represents critical generalization — pattern-developing — out of multiple ethnographic comparanda, often quite far-flung in origin ; and it offers the historian of religions an articulated and dynamic formulation for a kind of social system (Some classic models, of course, have seemed so intrinsically convincing that scholars assume their veracity without testing them against new data). The comparative use of such social models must then begin with the question, does the system it reveals apply to the scholar’s immediate data ? And as in the initial stages of the comparative enterprise, that data must first be conceptualized and described in comparative terms, and then studied as a potential exemplum of the anthropologist’s system. What data is missing from the historical record whose existence the model implies or requires ? In what ways might the data reflect the social model even while differing from the model’s classic exempla ? Can the ancient data provoke an adjusted form of the social model ? Douglas’s 1970 Natural Symbols, for example, which proposed a relationship between the symbols by which groups understood themselves, especially the body itself, and the degree of social articulation and social boundaries observed in the group, has had considerable influence on religious studies, especially studies of nascent Christianity, in offering a model to understand group behavior and, more importantly, to frame those behaviors comparatively.21

23It should be noted at this point that the critical use of anthropological models and ethnographic comparanda to enrich the description of religious context and justify the terminology by which ancient materials are translated and generalized does not require expertise in an entirely other world of data, any more than the anthropologist who develops a theoretical language of patterns on the basis of her fieldwork in Malaysia needs to do equivalent fieldwork in Guatemala, Nigeria, and Amsterdam in order to justify her models. To attend carefully and critically to the terminology and patterns used to describe and generalize ancient data, it is sufficient simply to look out for comparanda from other cultures and for dynamic models based on modern ethnography as resources for the sharpening and enrichment of the materials at hand. One researches the theme as it might be explored cross-culturally, not just the datum in its own historical context. That is, it is not infinitely broad expertise that is required to invite a comparandum but a sense of what it is one is trying to describe, to frame, to clarify.

  • 22 MacCormack 1981, 1991.
  • 23 MacCormack 1984, 2006, 2007.

24By way of contrast, one might turn to the eminent historian Sabine MacCormack, who managed over her career to master and publish in two entirely different historical fields, the late Roman empire and the colonial Inca empire, allowing her a rare capability to think across two distinct worlds of religious transformation and Christianization.22 However, the course of her oeuvre led not to the formulation of general models on the basis of two divergent fields of comparanda — the “illustrative” comparative mode explicated by Bonnell and Smith — but rather coincidental parallels (“much as x took place in late antiquity, so y took place in Peru”) and, in her most recent book, the proposal of an historical and genealogical relationship between notions of Roman empire and early modern notions of Inca and Christian empire.23

  • 24 Smith, 1990 ; 2000, p. 238.

25Now, this can be a risky endeavor as part of a comparative enterprise. The assertion of genealogical relations between comparanda, as Jonathan Z. Smith observed, has commonly concluded (or even driven theologically) comparative work in ancient Mediterranean religions : for example, mystery cults proposed as the source of Christian Eucharist, or Greco-Roman “divine men” as the source of Jesus as miracle-worker.24 Of course, genealogical relations may well constitute the only plausible conclusion from the relative historical contexts of the comparanda (especially if the scholar clarifies, as MacCormack does, the precise means of “influence”). But comparison in the service of deducing priority is an entirely different endeavor from what we have been exploring in this essay — closer to Bonnell’s model of analytic comparison, insofar as it examines the metamorphosis of ideas through the comparison of diachronically-arranged examples.

Comparison and/as Religious Studies : Ancient Data and the Conceptualization of Religion

  • 25 See, e.g., Poole, 1986, p. 429 – 430.

26Perhaps the development of larger themes and patterns, the illustrative comparative mode, presupposes a prior disciplinary commitment, less to the intrinsic interest and wonders of a particular historical period than to the implications of that historical period and its data for the broader understanding of religion. In my own case, the beginning and end of my historical work is the problem of the critical description of religion and religious dynamics, so I have a prior dedication to bringing the data back to the rectification of some larger theme or pattern. My primary interests lie therefore in general phenomena, like the mechanisms of religious violence, the mechanisms and domains of Christianization (as a subset of cultural transformation more generally — Hellenization, Romanization, Islamization), and the phenomenology of domestic and “popular” religion. My historical work in Roman Egypt and the late antique East serves as “fieldwork,” in the sense of those anthropologists who move between historically-contextual descriptions of cultures and the formulation of larger observations about, say, the function of demonology in popular discourse or the relationship of religion and landscape. Anthropologists think of their data as examples of something human, social, performative, responsive — worthy of framing for broader comparative purposes.25

  • 26 Bettini here above in this book.

27But historical work in religions inevitably involves combing through ancient data familiar and new in search of the aphormé, to borrow Maurizio Bettini’s term : that anomaly in text or archaeological record that serves as a point of departure, the datum that does not fit prior models or assumptions and thereby challenges our assumptions.26 My procedure then largely follows the method that Smith laid out : (a) to understand that anomaly as part of a system, (b) to describe it tentatively, (c) to ask of what phenomenon it might be an example, (d) to describe it according to some critical terminology,(e) to ask or seek out what might be a fuller or richer example ; (f) to articulate, on the basis of both proximate comparanda with limited context and distant comparanda with richer context, a plausible social context in which the datum makes sense, the anomaly familiarized — a context that may require abandonment or adjustment with further primary data ; and finally (g) to propose some broader observations about the pattern, the phenomenon, on the basis of this new data.

28By way of example I want to introduce two anomalies that served as points of departure for me, my initial efforts to contextualize them through proximal comparisons, and my recourse to social models and ethnography to gather a larger and more meaningful context for the data.

The Performative Construction of Demons and Spirit Possession27

  • 27 See Frankfurter, 2010.

29A number of authors from across the late antique Mediterranean world report that so-called demons would appear in Christian holy places and, before (or even instead of) being exorcised, would serve as oracles or mediums for assembled people. Most such reports come from the fifth and sixth centuries C.E., like this one by the historian Sozomen :

  • 28 Sozomen, Historia Ecclesiastica 7, 24, transl. Schaff, NPNF ser. 2, vol. 2, p. 393.

It is said that while the battle was being fought, a demoniac presented himself in the temple of God which is in the Hebdomos, where the emperor had engaged in prayer on starting out, and insulted John the Baptist, taunting him with having his head cut off, and shouted the following words: “You conquer me, and lay snares for my army.” The persons who happened to be on the spot, and who were waiting impatiently to learn some news of the war, were amazed, and wrote an account of it on the day that it occurred, and afterwards ascertained that it was the same day as that on which the battle had been fought.28

  • 29 [Firmilian to] Cyprian, Epistle 75, 10, 2 – 5 ; Athanasius, Festal Letter 42.

30What struck me about this and similar reports by Sulpicius Severus, Gregory of Tours, Athanasius of Alexandria, Shenoute of Atripe, and others was the ambiguity of the “demons.” The character of the reports suggested to me that this ambiguity was not simply some narrative trope but potentially revealed an ambiguity about the nature of possessing spirits in performance. Several authors — those closest to the events described — seemed themselves to waver over the nature of the spirits, referring to them sometimes as demons and sometimes in more neutral or positive terms.29

31What also struck me was the relationship of these demonic appearances, which sometimes culminated in the “demons’” acclamation of Christian divinities, and saints’ shrines. Did these possession performances reflect the resurgence of traditional gods in opposition to Christianity or a kind of local agency in the reception and acceptance of Christian authority ?

  • 30 This work was also stimulated by the works of Caciola, 2003, and Sluhovsky, 2007.

32Thus far I was working comparatively across a cluster of exempla from throughout the Mediterranean world and over several centuries : the kind of “comparison” that historians of late antiquity pursue habitually. However, the nature of my questions, my terminology (spirit possession, performance, local agency), and my tentative contextual models all depended on the richly contextual and theorized anthropology of spirit possession by Ioan M. Lewis (1989) and others. Pursuing the process of the local identification of possessing spirits more broadly in this literature, I learned that “demon” and its various equivalents are rarely (if ever) such static understandings of possessing spirits ; that, rather, possessing spirits that might be viewed by religious authorities in evil terms may offer devotees within the same religious system oracles, healing, and reassurance ; and that, even in historical Christian cultures more attentive to the control of spirits, the process of identifying a spirit as ancestral, prophetic, or demonic may last years, negotiated between seer, family, community, and various religious authorities.30 The larger pattern, then, of which the early Christian cases under discussion would be particular examples (along with many from medieval and early modern Europe and modern North Africa), involves the diachronic process by which spirits and spirit possession become interpreted and “situated” in communities, as a dialectical relationship between the mantic subject, the immediate community, and institutional authorities near and far. It also involves, I proposed, the means by which communities accept a new religious system, through the agency of possession.

33Comparison in this case went well beyond the mere collection of parallels, for more than anything else I sought through analogous cases the refinement of a terminology to describe phenomena “on the ground,” beyond the literary representation. Comparison also allowed the development of a model of Christianization that took into account possession, the discourse of demonology, and the novel space of the Christian shrine. And finally, it allowed the peculiar data for possession in late antique Christianity to contribute to a larger theoretical conversation about spirit possession and religious change.

Popular Christian violence in the generation after Christianization31

  • 31 Frankfurter, 2000b.
  • 32 E.g. Caseau, 2001 ; Sizgorich, 2009.

34Most discussion of violence in the Christianization of the late antique countryside revolves around the instigation of bishops and abbots, whereas villagers and laity appear (in primary and secondary materials) as either passive observers or paralyzed and anxious victims.32 Given popular participation in the destruction, or defense, of traditional shrines and images during the Christianization of Latin America, Polynesia, Africa, and other cultures, I have long wondered, would late antique villagers really have been so passive ?

  • 33 Besa, fr. 41 : “To the Dignitaries and People of the Village,” ed. & transl. Kuhn, 1956, 21, p. 129 (...)

35Near the end of the fifth century C.E., Abbot Besa of the White Monastery in Atripe in upper Egypt wrote to local villagers who had been, in his words, “fighting for nothing over a piece of wood.”33 In an era of increasing Christianization, following a period of iconoclastic violence under the prior abbot, Shenoute, this obscure allusion offered both an anomaly and a point of departure : what kind of wood would cause fighting ? What I proposed in a publication was that this “piece of wood” was actually an Egyptian divine image and that the violence over it, which Besa specifies as village-based, had arisen as a purge movement by local Christians against the keepers of such divine images : adherents of traditional Egyptian ritual expressions in a period when such expressions were less tolerated.

36I proceeded first by exploring the cultural and religious context of this region in Egypt in the fifth century, showing that by all accounts Christianity there and then was still in the process of being negotiated and assimilated. The language “only a piece of wood” was itself often used in Egyptian Christian polemic to dismiss the value of traditional statuary. Then I explored some religious situations elsewhere in Egypt from the same period that resembled the one I proposed for Besa’s village dispute. This was the first level of comparison : cases of local religious conflict in which traditional religious adherents — “pagans,” as it were — were subjected to violence on the part of Christian monks and/or villagers. This more elaborate documentation, I argued, filled out the historical picture of Besa’s village dispute.

  • 34 Douglas, 1999.

37But I also referred to a case that Mary Douglas had reported among the Lele of Zaire in the 1980s, in which local Catholic priests had incited a popular iconoclastic purge movement against traditional images and their elderly owners well after Catholic Christianity had been established in the region.34 My interest lay in the larger phenomenon of local peoples’ — as opposed to missionaries’ — “enactions” of Christian enthusiasm or exorcistic iconoclasm in the generations following the introduction of Christianity to a region. That is, Besa’s allusion to popular violence led me to hypothesize a pattern of popular iconoclastic purge (as opposed to that mobilized by bishops or abbots). Examples of just such popular iconoclastic violence following Christianization or reform from modern Africa, as well as early modern Europe and Mexico, showed me how the pattern might work and what contexts might be significant : the role of priests or lay leaders, the youthfulness of the participants, recent Christianization, the types of rumors circulating, even divergent relationships to landscape and buildings. My own comparative work thus consisted in developing the situation in Abbot Besa’s world as an example of post-missionary popular iconoclasm, and beyond that, religious violence of a particular sort.

38Now, it may seem as if I was simply producing a template and forcing historical data into it. But it is no different than proposing an hypothesis for anomalous ossuary or archaeological data — an hypothesis that can collapse or gain nuance with further data (or more apt comparisons) but ultimately allows one to make sense of artifacts and stimulate their discussion. At a broader level, hypotheses that propose artifacts or data as examples of larger patterns serve also the rectification of those patterns — and ultimately help us understand the phenomena of religion.

  • 35 As developed in such exemplary comparatively-informed studies as Friedland & Hecht, 1998, and Linco (...)

39Religious violence as a phenomenon, for example, only gains meaning as a category through particular incidences of iconoclasm, massacre, expulsion, incineration, desecration, or even (as some of my colleagues are currently pursuing) monastic discipline, which must be scrutinized in historical and social context and compared if one is at all to justify or constitute a category “religious violence.” Indeed, this case I described involves only one particular genus of religious violence, quite different from the destruction of the Bamiyan Buddhas, from Franciscan monks’ depredations in Mexico, Baruch Goldstein’s Purim massacre, or the demolition of the Ayodhya mosque for a Ram temple. As obvious or simple a term as it might sound, “religious violence” is itself a comparative pattern, one that implies — and should be defined in terms of — some dynamic system of action, mobilization, ideology, and religious discourse that can be experimentally applied to various historical cases in order to understand further their religious nature.35 Those cases to which I proposed the fifth-century Egyptian case could be productively (and carefully) compared, including modern African witch-cleansing and exorcistic movements, allowed some insight into how religious images and their owners might have been regarded by their persecutors in fifth-century Egypt. That is, the lingering, venerated religious image can appear as an obstruction to civic fortune to some, while to others of a more traditional bent the image may be the last preserver of civic fortune. But such insights arise from proposing the historical situation’s exemplification of a particular type of religious violence.

40My method thus involves looking at historical data with an eye for situations that might elucidate or complicate a pattern discussed comparatively in anthropological literature — or simply for anomalies, for aphormaí, that relate to general themes in the study of religion that have long interested me, like violence, possession, christianization, ritual, popular religion, and magical aspects of religion. It presumes, of course, wide reading in historical and ethnographic literature that addresses the kinds of religious phenomena that interest me so that I can recognize anomalies or problems of consequence, to describe them, and to have some sense of how comparanda manifest themselves cross-culturally. Finally, the method points in two ways : (a) to the critical reconstruction of an historical context, on the basis of comparanda, that renders the anomaly both familiar and exemplary ; and (b) to the greater understanding of some general pattern in the history of religions : e.g., the relationship of possession performance to religious change, the varieties of religious violence.


  • 36 See Smith, 1990.

41The kind of illustrative or taxonomic comparison described in this essay is not for everyone, and there is much work to be done in strictly philological, archaeological, or historical modes without the need to draw in other cultural examples or to read programmatically in history or ethnography outside one’s “home” field. However, few scholars in history, classics, or archaeology admit the degree to which the very language they use to translate and explore “paganism,” “sacrifice,” “magic,” “votive offering,” or “conversion” (inter alia) depends on comparison, on theoretical models and hypothetical patterns, whether they intend so or not. And if there is not some critical effort to lay out those models and implicit comparanda, to justify the term or the pattern, then the comparisons will inevitably come from those old theological notions of Catholic priesthood and liturgy that nineteenth-century scholars habitually projected onto their materials.36 So at one level comparison is intrinsic to the historian’s inevitable generalizations, and some measure of critical justification of one’s generalizations or terms, as ideas generated from a selection of examples — of comparanda — serves the integrity of historical work of any sort.

42But comparison may also involve, I have suggested, a more deliberate mode, in the formulation of a pattern — the holy man and his social function, post-mission popular Christian violence, spirit possession as an acknowledgment of Christian sainthood, and so on — that relates ancient data to richer ethnographic cases. And for those inclined to explore the data of ancient religions in this comparative mode, they must begin their explorations not with the question, “what is this datum like ?” but rather by asking, “Of what phenomenon or system in religion might this datum be an example ?” A pattern is articulated in rigorously “etic” terminology to maintain hermeneutical independence from “emic” language, even if it is only one or two particular cultures under discussion, and even if “emic” categories (like “wali” or “darshan”) might aid in the description of the pattern or lend it a short-hand label. More importantly, one uses the pattern not as a static grid to force on the data but as (a) a gauge of difference among cases across history while also a principle for relating those cases (how might Catholic statuary and traditional Egyptian wooden images be viewed differently by their zealous detractors ?) ; (b) a hypothetical dynamic system in which a text or archaeological datum makes sense pending additional data ; and (c) a means of bringing an historical artifact — text, object, event — into broader conversation with humanistic inquiry, not as an amusing anecdote or exotic anomaly but as a critically described case of some revealingly human phenomenon.


1 I am immensely grateful to Claude Calame and to the Casablanca Group for comments in the revision of this essay.

2 See, e.g., Luck, 2000, p. 223 – 38 ; Cameron, 2011, p. 3. See also Ascough, 2009.

3 See, e.g., Lane Fox, 1987 ; MacMullen, 1997.

4 Echoing Cameron, 2007, p. 43.

5 See Smith, 1982, p. xii – xiii ; 2004, p. 230 – 250. Examples : Smith, 1978, p. 147 – 171 ; 1982, p. 102 – 120.

6 See Smith, 1982, chap. 2.

7 See esp. Smith, 1990, p. 99.

8 Referring to the critique by Sandmel, 1962. See Smith, 1990, p. 51 – 53 ; 2000 ; and Paden, 2000.

9 Smith, 1990, p. 52 ; see also, 2004, chap. 8, esp. p. 194.

10 See esp. Smith, 1990.

11 On the category “sacrifice” see now Frankfurter, 2011.

12 Many of the criticisms Claude Calame marshals against anthropological interpretation (2002) are all the more applicable to the less self-conscious disciplines of ancient history and archaeology.

13 See, e.g., Dodds, 1965 ; Lane Fox, 1987 ; Stroumsa, 2009.

14 See Smith, 1978, p. 172 – 189 ; Frankfurter, 1998, p. 198 – 237 ; 2000a.

15 See, e.g., Frankfurter, 1998, p. 131 – 142.

16 Brown, 1970, following Douglas, 1966, p. 94 – 113. See also Brown, 1978, chap. 2, and Brown, 1998, esp. p. 359 – 360.

17 E.g. Brown, 1982, p. 118, 123, 124, 131.

18 Brown, 1978, chap. 3.

19 Brown, 1998, p. 303.

20 See, e.g., Kaplan, 1984 ; Stirrat, 1992 ; Howard-Johnston & Hayward, 1999 ; Frankfurter, 2003.

21 E.g. Gager, 1982 ; Meeks, 1983 ; Kraemer, 1992.

22 MacCormack 1981, 1991.

23 MacCormack 1984, 2006, 2007.

24 Smith, 1990 ; 2000, p. 238.

25 See, e.g., Poole, 1986, p. 429 – 430.

26 Bettini here above in this book.

27 See Frankfurter, 2010.

28 Sozomen, Historia Ecclesiastica 7, 24, transl. Schaff, NPNF ser. 2, vol. 2, p. 393.

29 [Firmilian to] Cyprian, Epistle 75, 10, 2 – 5 ; Athanasius, Festal Letter 42.

30 This work was also stimulated by the works of Caciola, 2003, and Sluhovsky, 2007.

31 Frankfurter, 2000b.

32 E.g. Caseau, 2001 ; Sizgorich, 2009.

33 Besa, fr. 41 : “To the Dignitaries and People of the Village,” ed. & transl. Kuhn, 1956, 21, p. 129 – 130 (text), 22, p. 123 – 124 (transl.).

34 Douglas, 1999.

35 As developed in such exemplary comparatively-informed studies as Friedland & Hecht, 1998, and Lincoln, 2003.

36 See Smith, 1990.


Boston University (Department of Religion)

Le texte et les autres éléments (illustrations, fichiers annexes importés) sont sous Licence OpenEdition Books, sauf mention contraire.

Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search