Studying the "folklore" of the Middle Ages is a frustrating enterprise. Our sources yield a harvest rich enough to whet the appetite yet still too slight to satisfy. From the early Middle Ages, law codes and penitentials depict healing rituals in which children were placed in ovens, fairies placated by casting bows and arrows into barns, and unbaptised children who were "staked" to stop them rising from the grave. From the late twelfth century onwards, exempla indicate that some men and women thought it unlucky to meet a priest in the street and that others were not averse to crumbling communion wafers over their crops to protect them.

The twelfth and thirteenth centuries witnessed a proliferation and diversification of "historical" writing. Chroniclers increasingly found space in their narratives for tales of the wondrous: Ralph of Coggeshall told of green children found in cornfields, wild men fished from the sea and invisible spirits haunting peasant houses (Ralph of Coggeshall 1875, 117-21). On the fringe of the chronicle genre, new species of narrative also evolved that set the wondrous and fantastical at the centre rather than the edge of the historical enterprise: here Walter Map told tales of fairy women who married mortal men and stories of dead men who rose from the grave by night (Map 1983, 154-6 and 344).

It is easy enough to assemble examples, but how are we to analyse them? In handling such material, three problems are immediately visible. (In a sense they are a single problem, but for the purposes of clarity it might be wise to split them at the outset.) The first is conceptual: can we speak of "folklore" in the Middle Ages and, if so, who exactly were "the folk" who used the lore? The second is evidential: the communities of medieval Europe have, because their cultures tended to be articulated by oral rather than written forms, left only the very faintest traces of beliefs and practices.

These seldom survive in sufficient concentrations to allow us to describe the beliefs and practices of any single community or even a particular region. This closes down possible avenues of exploration: the opportunities for "micro-history" or "thick descriptions" are few in this period, especially in the earlier part of it (the obvious exception here is Le Roy Ladurie's Montaillou [1980]). A final complexity is related to the second but is methodological in character. Our scattered fragments of evidence are products of clerical pens, of a literate culture rather than of the predominantly oral culture in which beliefs were held and practices used. As such, the cultural gaps between the practitioners of oral and written culture call for thought.

This problematic trinity--a three-in-one historiographical conundrum--demands that we deploy powerful theory if we are to recreate from fragmentary remains the larger patterns of medieval belief and practice. In discussing the possibilities here, we cannot speak of "folklore" in isolation or as a given category. Rather, we need to think of the bigger picture, considering "religious culture" as an organic whole.

One series of solutions to our problems is supplied by "popular/elite" models of medieval religion. Over the past twenty-five years, it has often been argued that religious culture in medieval Europe can best be understood in terms of social categories, that considerable gulfs existed between religions of educated elites and uneducated masses or between those of clergy and laity. Jean Delumeau, in challenging unreflective images of the Middle Ages as an "Age of Faith," questioned the comprehensiveness of the conversion process (Delumeau 1971).

He argued that, although social elites may have been converted to full-blooded Christianity, the masses were only superficially Christianised and continued, beneath a veneer of official observance, to practice a protean, pagan folk-religion. This religion was articulated orally and so left few traces, occasionally preserved by the pen of elite churchmen who, more often than not, noted it for the purposes of repression. This hard-edged analysis that seemed to over-correct the older orthodoxy has been softened by later generations of scholars (Schmitt 1983; Gurevich 1988, xiii-xvi; Le Goff 1980; 1988). Jacques Le Goff and his disciples suggest that "elite/clerical" and "mass/folkloric" cultures were distinct but in a state of constant dialogue, churchmen seeking to modify the latter by a mixture of repression and reinvention (Schmitt 1976, 941-53).

In some ways this model is beguiling. In allowing us to distinguish the official Christianity of the elite from the popular religion of the folk, it provides a ring-fenced folkloric space in which we can situate accounts of green children, fairy wives and spectral knights. Yet even the more refined and nuanced approaches of Le Goff and Schmitt fail to convince fully. Criticisms of their work concentrate on the assumption that social categories such as elite/mass or clerical/lay are good predictors of religious beliefs (Delaruelle 1975; Murray 1978, 14-17, 237-41, 244-57 and 319; Brown 1981, 18-20; Brooke and Brooke 1984, 9-10 and 12-13; Rubin 1991, 7; Tellenbach 1993, 128). A cursory examination reveals that medieval society was sufficiently complex for many people to evade such classifications. Where, for instance, does the parish priest fit in? Is he a subscriber to elite or popular religion? He certainly falls into the clerical category, but how meaningful is this label? Local priests were drawn largely from the peasant communities they served and were probably afforded only rudimentary education. The difficulties multiply if we think of noble monastic converts and peasant hermits. Were their beliefs necessarily different because of their different backgrounds, or might they have shared an ascetic, world-rejecting spirituality? If we turn to praxis, we run into similar difficulties because we can find many practices that transcended social categories. Casting our eyes across the channel for a moment, we find that the penitential handbook written around the year 1000 by Burchard of Worms, the Corrector, condemns those who feared to go outside before cockcrow because evil spirits were abroad (Burchard of Worms 1898, 442). This might be read as an instance of the beliefs of a literate clergyman who participated in elite culture being in tension with the folkloric beliefs of the masses. But how then do we account for Guibert of Nogent, an educated churchman of the twelfth century who was accustomed to sleep with a lamp by his bed to keep evil spirits at bay (Guibert of Nogent 1970)?

Normative and Narrative Sources

The temptation to categorise belief on the basis of social groups such as elite and masses, clergy and laity, literate and illiterate is born, I think, of the kinds of evidence historians have dwelt upon and the language they have absorbed from them. Much of the work of Le Goff and Schmitt has emphasised collections of sermon stories, penitential handbooks and canon law texts. Such sources have an explicit didactic project and they make a sharp contrastive distinction between Church ideals and malpractices in wider society. But have such contrasts in the normative sources been taken too much at face value? The works of a twelfth-century churchman, Gerald of Wales, offer an opportunity to test the hypothesis because here we have an author who wrote about similar issues of belief and practice in different literary genres.

Gerald, born in about 1146, was well educated and had spent time in the schools of Paris before returning to his native Wales to become archdeacon of Saint David's (Bartlett 1982, 27-57). One theme treated several times in his voluminous writings was whether laymen and women should be allowed to dance in churchyards on feast days. It emerges in his Jewel of the Church, a collection of instructional stories that readily betrays a school training in canon law. Here Gerald promoted a vigorous reforming agenda, attacking a wide range of abuses in the wider community. In one story, Gerald warned "that the laity ought not to engage in singing and dancing around churches and cemeteries on the feasts of saints but should devote themselves to the divine service" (Gerald of Wales 1861-91, vol. 2, 119-20). Gerald buttressed his argument with references to the Council of Toledo and the authority of the Fathers, specifically Augustine, and the view he expressed also chimed with those of countless earlier and later penitential manuals (see, for example, Morey 1937, 256). Yet Gerald's view of such dancing was not so straightforward as the Jewel of the Church might imply. He also visited the subject in a very different kind of writing, a narrative account of a preaching tour he had undertaken through Wales in 1188, the so-called Journey Through Wales. Here Gerald described how, at Saint Eluned's church in Brecon, on the feast day of the saint, all the people of the neighbourhood congregated at the Church and then danced around the churchyard, singing "traditional songs" (Gerald of Wales 1861-91, vol. 6, 32-3). Some of them fell into frenzy and began to mime work they had illicitly performed on feast days--"you might see this man imitate a cobbler, that man a tanner." Having done this, they were led to the altar, made oblations there, and "returned to themselves."

Here, it seems, we find an unofficial ritual dance serving penitential purposes in a community where Church teaching about abstaining from work on feast days was well established but where official rituals for expiating sin were not. Hence the community shaped its own redemptive strategies. More interesting for our purpose, however, is the reaction of archdeacon Gerald. Given the views expressed in his Jewel of the Church, we might reasonably expect him to condemn such unofficial practices. In fact, his reaction was strikingly different: "God desires not the death of a sinner, but rather that he may turn from wickedness: and so, by taking part in these festivities, many at once see and feel in their hearts the remission of their sins, and are absolved and pardoned" (Gerald of Wales 1861-91, vol. 6, 33). An archdeacon, whose theoretical role was to control and limit diversity of religious expression, was here, in practice, sympathetic to it.

What moral might we draw from this story? Normative sources may give us insights into official Christian teachings, but they do not necessarily give us straightforward access to the religious beliefs of churchmen, not even the particular churchmen who wrote them. Gerald's view of dancing in churchyards seems to change depending on which genre he is writing in. When producing an instructional work, Gerald felt constrained by canon law and the authoritative utterances of Church Fathers and Councils. But when close attention to these authorities was less urgent, as in the Journey through Wales, Gerald's beliefs emerged in a different shape.

This example suggests that there are risks in viewing both exempla and handbooks of penance through the pre-formed lenses of "popular" and "elite." Approaches to medieval religion that rely on such distinctions are problematic because they confer on "popular" or "folkloric" belief unconvincing unities, and accentuate the tensions between it and putative elite religious culture. Such readings encourage, and are encouraged by, analyses that are openly or tacitly attached to ideas of antagonistic, class-based social relations. These approaches can be restrictive. They make our evidence the product of an elite that, when faced by mass culture, was at best deeply unsympathetic, at worst hostile and uncomprehending. No one would argue that exempla give unproblematic access to the culture(s) of the masses, but they are often treated as if they provide quite straightforward insights into those of the clerical elite (see the critical comments in O'Neill [1986, 222]). Authors of normative sources such as handbooks of penance or sermon materials operated self-consciously within literary and theological traditions. During the process of composition, their theological and canon-legal training found forceful and articulate expression, inducing them to draw heavily on the canon of scriptural, patristic and later texts that would give their own writings authority. The act of disseminating official Christian teachings imposed such a discipline. But it does not follow that the religious and cultural values of such authors were formed solely in schools or cloister, shaped by canon law and theology alone. Their religious and cultural make-up was also moulded, to a greater or lesser extent, by other "local" cultural traditions in which they participated perhaps during childhood, through pastoral ministry in later life or even through excursions from the cloister into the world. The key question here is what happens when clerical authors are freed from the constraints of normative genres. If we read only penitentials and sermon literature the question is an unanswerable one, but if we turn to narrative writings we get a chance to observe clerical authors (sometimes the same clerical authors who produced our normative sources) in a different literary light, governed by different "rules" of genre. Hence, in thinking about evidence for medieval religious culture, it is important to consider not only who the author of that evidence is, but also the context or genre in which he was writing.

Christian Conversions and Pagan Survivals

Contextualisation is also important in a second sense. As we have observed, sources for belief and practice in the Middle Ages are comparatively thin on the ground. This has tempted some historians to expand the geographical scope of their studies, often examining religious culture on European or still larger scales. Historians and folklorists interested in tracing "pagan survivals" from the pre-Christian into the Christian era have been especially inclined to this approach. Most recently, Carlo Ginzburg has used it to argue that the hard core of the religion of the folk was scarcely altered by Christian conversion and that we can discern fundamental continuities of belief running from antiquity through the Middle Ages into the early modern period (Ginzburg 1990). Yet there are many dangers in such a method, not least that in considering examples out of context, astonishing cultural continuities over space and time dazzle the observer and obscure fundamental differences.

Such suspicions are reinforced by a closer look at the evidence from England and Scotland in the central and late Middle Ages. This reveals few compelling instances of pagan continuity. One of the strongest cases could be made for practices described in Reginald of Durham's Book of the Wonderful Miracles of Blessed Cuthbert. He observed how Abbot Ailred of Rievaulx had been scandalised during a visit to Kirkcudbright, Galloway, by the ritual sacrifice of a bull performed by members of the local community (Reginald of Durham 1835, 179). On the face of it, this looks like a classic pagan animal sacrifice such as those widely condemned by ecclesiastical authorities during and immediately after the Christian conversions. Yet on closer inspection, this ritual is more complex. It was orchestrated by so-called "scholars" (or "scollofthes" according to the text) from the religious house of Kirkcudbright and, most interestingly, the bull was offered to Saint Cuthbert rather than to any pagan deity. In other words, outward ritual forms that appear to be pagan were, in fact, articulating devotions to a well-established Christian saint.

The temptation here is to separate Christian from pagan, breaking the narrative of the ritual into fragments and tracing the lineage of each back through time. Yet the complexities of this account suggest that this temptation should be resisted. Rather than seeking to label each facet of the ritual as "Christian" or "pagan," we should study it as an organic whole in order to get some sense of its function and how it was understood by members of the community. Searches over great tracts of space and time for "morphologically" similar examples are unlikely to yield answers to these kinds of questions (Purkiss 2000, 7). Patrick Geary has pointed out that medieval religious culture was intimately interconnected with other dimensions of life in local communities, with agriculture, social hierarchy, family, law, politics (Geary 1994, 32-3). If we accept this idea of "interconnectedness," future research might, where permit, look more deeply into the culture of particular localities and regions. By putting wider aspects of daily life under the historian's magnifying lens, we can get a sharper sense of the larger-belief system in which particular rituals or practices were located.

Local Religious Cultures

Such conjectures do not solve the problem of how religious culture in these communities should be related to that of a universal Church. To do this, we need to rethink the larger theory and move away from narrow notions of "popular" and "elite." Here a modified version of Peter Burke's model of early modern culture can help us. Burke distinguished two cultural traditions--a "great tradition" fostered through literate modes in schools and universities, and a "little tradition" that was locally various and sustained by oral means among the unlettered inhabitants of local communities (Burke 1978, 22-8). These traditions were not hermetically sealed: although the masses were essentially cut off from the "great" tradition, the elite were able to participate at will in the "little" tradition.

This model cannot be applied as it stands to medieval religious culture. It is true that substantial bodies of complex theological learning existed in monasteries and in the schools, and a great part of this never percolated into the parishes in even the most rudimentary forms. Yet the institutions of the Church did not set out to preserve the articles of religious belief in the same way that grammar schools and universities sustained Burke's elite culture. Christianity was not an exclusive tradition. Rather the reverse, monasteries, cathedral schools and, more especially, the structures of the Church at diocesan and parish level were conceived, to varying degrees, as means by which Christian belief and practice were to be spread on the world. R. I. Moore has argued that preaching and teaching, especially by the parish priest, was designed to bridge any gap that might exist between the position of the Church on one hand and belief and practice in local communities on the other (Moore 1994, 35). Such impulses served to break down cultural boundaries. There was, for example, no clear-cut distinction between oral and written cultures. The oral relations of the peasant could be, and periodically were, captured in script by chroniclers, authors of miracles collections and others. More significantly for the spread of Church teaching, versions of the "written culture" of the Church were made accessible to the peasant through recitation and preaching from the pulpit, what D. H. Green calls "diagonal channels of communication" (Green 1994, 169-72).

Such insights lead us to an adapted version of the Burke model. We can distinguish two traditions in medieval religious culture but all members of medieval society participated to a greater or lesser degree in both of them. This, essentially, is the upshot of work by W. A. Christian on sixteenth-century Spanish religion. He has stressed the "placebound" character of many beliefs and observances, and draws a distinction between "universal religion" based on sacraments, liturgy and sacred time and local religion rooted in shrines, images, saints and relics (Christian 1981a; 1981b; O'Neill 1986, 222-5). This local religion provided a range of preventive and remedial strategies ranging from processions, votive masses and invocations of special saints to searches for signs in the weather and prognosticatory lotteries, all shaped to the needs of the immediate community.

These findings can guide us when thinking about the Middle Ages. The systematic teachings of the Church can be picked out with some confidence in this period, and we can perhaps label these "official Christianity." Such teachings mingled in the localities with fluid and locally varied beliefs that did not grow directly out of the systematic teaching of the Church. These beliefs existed within the interstices of official faith and ritual and churchmen did not necessarily see them as pagan, unchristian, heretical or erroneous. We might be tempted to describe these beliefs as "folklore," but it might be safer to call them "unofficial beliefs" because they were not simply the property of "the folk" (a term that, I think, encourages us to see beliefs of the "masses" as easily detached from those of "elites"), but also of many educated churchmen as well. The whole notion of "the folk" and their "lore" becomes troublesome here. Although the line between official and unofficial might seem clear to us, armed with canon law tracts, sermon books and penitentials, it was probably not so clear to the inhabitants of local communities. Nor can we assume that these lines were clear-cut even for the diocesan clergy, let alone those of the parish.

None of this excludes the potential for tensions within the system of belief. Returning to a familiar example, if we re-examine the bull sacrifice at Kirkcudbright, we can begin to distinguish subscribers to different cultural values and the frictions that emerged between them. We have on one hand our author Reginald, a Benedictine monk from Durham. He was a friend of Ailred of Rievaulx, who had travelled up to Galloway to visit a Cistercian house at Dundrennan. En route Ailred had run into the ritual bull sacrifice performed by the "scholars" of Kirkcudbright. In this account, we find on one hand the austere spirituality of the enclosed Cistercians rooted in credal Christianity and strict obedience to their Rule; a culture in which Abbot Ailred had been deeply immersed since his early twenties. On the other hand, we find a community of independent "scholars," monks not tied to any religious order, free to practice distinctive local rituals and heavily involved in the religious life of the lay community living around them. Tensions between these two religious cultures were sharpened by ethnic distinctiveness. Walter Daniel, another Cistercian and biographer of Abbot Ailred, characterised Galloway as "a wild country where the inhabitants are like beasts and altogether barbarous" and noted how Rievaulx had "planted" the abbey of Dundrennan amidst the savagery to spread the Cistercian message (Daniel 1950, 45-6). Here then, we have a sharp encounter between the ardent official Christianity of the Cistercians and the localised, syncretistic Christianity of a community from the upland peripheries. Yet the encounter is sharp precisely because it involves the juxtaposition of especially (and unusually) concentrated forms of official and unofficial belief.

If we move away from this coincidence of extremes to Gerald of Wales and the rites in Saint Eluned's churchyard, a different picture emerges. Here we have not a senior and zealous monk in a new reforming order, but a secular clergyman, an archdeacon, responsible for pastoral care in the marches. Gerald was well acquainted with canon law and might make authoritative utterances about dancing in churchyards when writing instruction manuals, but when he met diverse practices on the ground he was more tolerant. Gerald lacked the resources to impose Church law by force. As a churchman working in the community he had to negotiate official Christian values and those of the communities he encountered. Inevitably, there must have been cultural compromise in such situations, and here we seem to see it. Gerald was interested in the core elements of the Church's reform agenda such as lay control over churches and enforcing clerical celibacy (Bartlett 1982, 29-31). Where he found novel ritual articulations of official Christian ideas, his attitude was more relaxed. Such pragmatism may also have been informed by discriminating sympathy. It is conjectural but nonetheless possible that Gerald's Cambro-Norman upbringing and career in the Marches may have soaked him in the culture of these places and hence inclined him to see these rituals in a more positive light.

This is, admittedly, slender evidence on which to base general conclusions. If, nonetheless, we pursue this line for a moment, we can perhaps still say something about method. Three things emerge here, all of which are, in one shape or form, arguments for greater contextualisation. First, we need to think of medieval religious culture as a commingling of unofficial and official belief that varies over space and time--forming not a series of cultural compartments, but a spectrum. This spectrum comprehends, albeit at the opposite ends, the beliefs of peasant and prelate alike (Van Engen 1986, 519-52; Duffy 1992). Second, it is dangerous to fence "folklore" off as a special object of study, and still more dangerous to associate the term with particular social groups. Rather we should regard the accounts that we tend to label "folklore" as facets of larger cultural systems that bind communities. Finally, we need to study accounts of belief and practice in authorial context, considering the impact of genre, authorial objectives and, perhaps most importantly, the cultural formation of the author as well as that of the community he described.