The Celtic New Year and Feast of the Dead

Abstract

At the opening of the twenty-first century it was an established orthodoxy in books about ‘the pagan Celts’ that they celebrated their New Year at the feast known in Irish as Samhain, which later became All Saints’ Day, and that this was also their great annual commemoration of the dead. This belief can be traced to the influence of two eminent Victorian and Edwardian scholars, Sir John Rhys and Sir James Frazer, although the part regarding the New Year had older origins. This article is intended to consider this belief’s development and the evidence that those who expounded it provided to support it. The article will conclude by widening the investigation to consider whether any other sources may exist to give credence to the idea.

Introduction

At the opening of the twenty-first century it was an established orthodoxy among respected scholars who published popular books on ‘the pagan Celts’ that they had celebrated the opening of each new year on the date which was 1 November in the Roman calendar, later the Christian feast of All Saints. Its morrow became another great ecclesiastical festival, of All Souls, and its eve the popular one of Hallowe’en. In the early 1990s a classic work on the ancient Celts by an expert from the previous generation, Nora Chadwick, was reissued unrevised, which called 1 November ‘the beginning of the Celtic year, at which time any barriers between man and the supernatural were lowered’ (Chadwick 1991, 181). Soon after, Hilda Ellis Davidson, a leader of studies into ancient northern European religion, declared that the most important annual festival of that religion had been that ‘marking the new year, celebrated at the end of October, at the beginning of winter’ (Davidson 1993, 88). In the middle of the decade, a pioneer of research into pagan Celtic Britain, Anne Ross, announced that ‘the Celtic New Year began on the eve of 1 November’ (Ross 1995, 434). In the later part of the decade, the most distinguished authority on the archaeology of the British Iron Age, Barry Cunliffe, agreed that the new year of that culture began on 1 November, and that ‘it was a liminal time between the two years and as such was dangerous: the spirits of the dead could roam free … The strength of tradition which lay behind this festival has ensured its survival as Hallowe’en and its Christianized version, the festival of All Souls’ (Cunliffe 1997, 189). His comments brought in two associated themes that were also commonly found in books of this kind: that the pre-Christian feast concerned had been especially associated with the dead, and that it had given rise directly to the Christian festival of All Souls on 2 November. In the first years of the new century, the entry on the preceding feast of All Saints on 1 November in The New Catholic Encyclopedia declared that it had descended from ‘a Druid festival of the dead’ (Smith 2003, 290).

None of these authors provided any evidence or source references for these statements: they were repeating something so well known and established that it apparently required no substantiation. It is also notable that all of the British scholars cited made, in further comment on the matter, the Irish feast known in medieval texts as Samain—or in late Irish orthography as Samhain—the model for their portrait of this new-year festival throughout the ancient Celtic world. A further common feature between these scholars is that they all took as the basis for their portrait of that world the archaeological evidence from the pre-Roman West European Iron Age, interpreted at times with the aid of medieval Irish and Welsh literature. By the time at which they wrote, the belief in a ‘pagan Celtic New Year’ and feast of the dead at the opening of November was indeed mainly concentrated among specialists in the Iron Age. It was not expressed by most experts in medieval Irish and Welsh literature, or most of those in the history of the Christian ritual year: the entry in the encyclopedia cited earlier was a recent, and very rare, exception. Since the beginning of the century, specialists in the Iron Age, especially in Britain, have generally stopped writing popular books on the ancient Celts, because many of them have lost faith in the concept of a unitary ethnic, linguistic, and cultural province inhabited by such people, and the term ‘Celtic’ itself had become controversial for ancient Europe (see especially Chapman 1992; S. James 1999; Carr and Stoddart 2002; Collis 2003).

This development, however, left the field clear for books which embodied a more traditional view of ‘the Celts’, like those cited earlier, to continue to circulate and determine the views of the general public. In the case of the notion of the Celtic New Year and feast of the dead, a quick trawl of current Internet websites serves to show that belief in it is still very much alive in international popular culture. In the mid-1990s I directly questioned the soundness of the evidence for it, but did so only to a brief, and so inevitably superficial, extent (Hutton 1996, 363–65). Moreover, I returned an open verdict on its actual verity. No apparent scholarly discussion has resulted, and so now, almost thirty years later, it may be timely to look in more detail at that evidence, and the issues that it raises.

The Origins of the Construct

The intellectual trail that led to the orthodoxy of an ancient Celtic New Year at the opening of November seems to have commenced in 1847 with an expert in medieval Irish literature, John O’Donovan. He did so with reference to a specific text, an entry in Sanas Cormaic, ‘Cormac’s Glossary’. One word defined in it was fogamur, which remains (usually as fómhar) the normal Irish word for autumn, and O’Donovan translated the entry as ‘Foghamhar, was given as a name to the last month’. He therefore concluded that the calendar year had ended with it, while adding the caveat ‘if there be no error of transcribers’ (O’Donovan 1847, lv). His suggestion attracted little attention for over forty years, until it was taken up and given wide publicity by the eminent philologist Sir John Rhys, in a high-profile series of lectures delivered in 1888 and published in 1892.¹ His starting point was Sanas Cormaic, but he hit a problem with it—that in the intervening decades a scholarly edition of the text had been published, by Whitley Stokes, which rendered the vital entry as: ‘Fogamur it is a name for the last month in autumn’ (Stokes 1862, 20–22). Rhys correctly pointed out that O’Donovan and Stokes had used different versions of the glossary. That used by O’Donovan was earlier. It seemed to Rhys literally to read, ‘Fogamur, for the last month it is a name in the autumn’, which he corrected to mean, ‘Fogamur: for the last month that is a name’. He further supported his case by pointing out that the last night of October, Hallowe’en, had been a great time for divination and the unveiling of mysteries in both Ireland and Wales, as befitted a turn of the year. In Wales until recently, women had gathered in the parish churches for rites to find who would die in the next year. He also noted an Irish tradition that on Samhain a bonfire had been lit at a place called Tlachtga, from which all the fires of Ireland were rekindled, which seemed another new year rite to him; and he noted that bonfires were known in Wales also, until recent years, at Hallowe’en (Rhys 1892, 513–15). This first appearance of his argument for a distinctive Celtic New Year rapidly found an audience: for example, it was repeated, although as not quite proven, by a clerical antiquarian, J. Fisher, in 1895 (Fisher 1895, 104–105).

Rhys returned to the subject with fresh material in 1901, as part of one of his most popular books. That material consisted entirely of folklore that he had collected on visits to the Isle of Man. He noted that on the island, land tenures and servant contracts commonly terminated at the end of October, as if at the end of a year. He had also found a reference in John Kelly’s Manx–English Dictionary, written in about 1808, that Manx mummers at that time opened their play at Hallowe’en with the words, ‘Tonight is New Year’s Night’ (Kelly 1858, 24). Kelly used this fact to become, apparently, the first person to suggest that Celtic peoples might have celebrated their new year then, but he had no impact because his book was not published until 1858. His suggestion was then apparently ignored until Rhys discovered it. The mummers no longer used that line when Rhys visited the island, but to his delight he found an old man who related that in his youth, in the early nineteenth century, he had served a farmer who told his hired hands that Hallowe’en had once been New Year’s Eve, which seemed logical to the farmer because of the termination of tenures and contracts then. Rhys then reported that the large number of divination rites held on that evening, to tell who would marry, or be born, or die in the coming year, was also a feature of Man, and supported his theory of a Celtic New Year held then (Rhys 1901, 315–20). By 1905 he could refer confidently to ‘the first of November, to which Celtic folklore unanimously points as the calends of winter and the beginning of the year’, as a proven fact, requiring no further argument or evidence (Rhys 1905–1906, 8).

In using evidence in this manner, Rhys was tapping into two major scholarly developments of his age. One was the concept of a pan-Celtic cultural province in ancient Europe, covering most of the west of the continent, with common language forms, customs, and deities. This was being promoted by French academics such as Henri d’Arbois de Jubainville, and appealed also to Irish and Welsh nationalists, and Rhys was one of the latter. The other was the belief that modern folk customs frequently represented survivals of ancient religion, fossilized in an unchanging rural culture, and as such could be used to reconstruct prehistoric beliefs and rites. It had developed among German researchers, such as Wilhelm Mannhardt, and had been taken up enthusiastically by British colleagues near the end of the nineteenth century. Both fashionable new ideas heavily influenced Rhys’s use of evidence in making his contention.

That contention was repeated, and given even wider currency, by another very distinguished academic, the classicist and theorist of comparative religion Sir James Frazer. He did so in 1905, in connection with his own interest in worldwide feasts of the dead, and fully amplified his case two years later, in a second edition of the same work (Frazer 1907, 301–18). Like Rhys, in forming his theories he was influenced by a predecessor, in his case Sir Edward Tylor, one of the British founders of the discipline of anthropology, who thirty years before had noted that across the world human societies held rites to commemorate and commune with their ancestral dead. Tylor linked this to the Christian reverence for martyrs, and festivals to honour the departed, such as All Souls’ Day, although he noted that around the globe such festivals were held in different cultures at a range of seasons. He held that the institution of All Souls’ Day by the Western Church had given a new lease of life to this ancient tradition, and, with all the contempt of a rationalist brought up in Protestant tradition, he sneered at the manner in which Catholic nations celebrated it ‘with relics of savage animism scarcely to be surpassed in Africa or the South Sea Islands’ (Tylor 1871, 32–38). Frazer fully shared Tylor’s rationalism, and contempt for superstition and Catholicism, and extended it to a general desire to discredit religion, and especially Christianity, in modern society (Ackerman 1987).

He began his discussion of feasts of the dead, as Tylor had done, with a worldwide survey of them, but placed a particular emphasis on those which were combined in some societies with a celebration of the New Year. These were concentrated in South-East Asia, but he also drew attention to the number of annual commemorations of dead persons—representing more than half of his sample of extra-European societies, and scattered across three continents—which did so at the end of harvest or the beginning of winter. He considered European observations linked to the festival of All Souls and concluded that a comparison of them with those held across the world ‘can leave no room for doubt’ that ‘this festival is nothing but an old pagan feast of the dead which the Christian, unable or unwilling to suppress, resolved from reasons of policy to connive at’ (Frazer 1907, 315). This is where Rhys’s argument came in useful, because if All Souls had fallen at the Celtic New Year, and rites to honour the dead were commonly held by humans at that time, it strengthened Frazer’s contention. He noted that the festival had been instituted by Odilo, abbot of Cluny, at the end of the tenth century, and spread outward from the Cluniac network of monasteries. Frazer concluded that an old Celtic commemoration of the dead at the opening of November must have lingered in France, a Celtic country, and eventually been incorporated into Christianity as ‘a concession to ineradicable paganism’ (Frazer 1907, 316).

He reinforced this sequence of suggestions by noting that the institution of All Souls had been preceded by that of a different feast, of All Saints, on 1 November, and adding that this had probably represented an earlier attempt by the Church to suppress the Celtic rites to the dead. Its plan was to substitute the saints for the dead as the object of veneration. He observed that this feast had been established by the Carolingian emperor Louis the Pious in 835 at the behest of Pope Gregory IV, who may have been trying to eradicate a pagan Frankish custom. However, the writings of the English monk Bede showed that such a feast had already been established at his time, the early eighth century, in England, which Frazer called another (fundamentally) Celtic country (Frazer 1907, 317). When this attempt proved a failure, Frazer continued, the Church had to let in the unsaintly dead, and as the saints had got 1 November, the rest of the dead had to get the following day. All of this was a double snub to medieval Catholicism, suggesting that not only was it riddled with pagan influences, but that it had been too weak in its hold on the minds of the laity to resist them. More than ten years later, Frazer gave fresh support to Rhys’s contention that this had been the time of the Celtic New Year, by repeating and endorsing two of Rhys’s pieces of evidence: the number of divination rituals held then, and the story of the relighting of all the fires of Ireland from a sacred one at Tlachtga. He added that the number of ceremonies associated with the dead at that time, across Europe, bolstered his own conclusion that the New Year feast had been dedicated to them (Frazer 1919, 224–28).

These ideas were published in successive volumes of his enormously influential masterwork, The Golden Bough, and summarized in the one-volume popular edition of it which was the one read by most of the general public (Frazer 1925, 633–34). The joint impact made by Frazer and Rhys had already been considerable. In 1911, the view of Samhain as the opening of the ‘Celtic year’ was repeated as established fact by John MacCulloch, a folklorist and author on the ancient Celts (MacCulloch 1911, 265–66). In the same year another widely-read folklorist, the American Walter Evans-Wentz, repeated with the same confidence that it had been ‘the great Celtic feast of the dead’ (Evans-Wentz 1911, 439). It was the combined influence of all four writers, but especially Frazer, that lodged both beliefs so firmly in the modern mind that they could be proclaimed at the other end of the century, as shown, without need for any supporting detail. In between, the line of transmission was kept up by their appearance, again as proven scholarship, in popular works such as those of the British scholar of comparative religion, Edwin James (E. O. James 1961, 316–17).

The Original Evidence

In evaluating the strength of the double construct, it may be best to begin with the evidence produced to support it when first formulated. The very first piece, which inspired Rhys, is the entry in Sanas Cormaic. This work still lacks a complete and detailed investigation.² It was allegedly first compiled by a king and bishop of Cashel in Munster, who lived around the year 900, but the entries were clearly multiplied and amplified over a long period, possibly extending into the later Middle Ages. The material was only put into alphabetical order at a subsequent stage, and much re-edited to put different explanations under a single headword (Russell 1988). For this reason, the most recent scholar to investigate it as a source for the nature of pre-Christian festivals has firmly rejected its reliability in this respect (Williams 2005). Of the two variants of the particular entry relevant here, that edited by Whitley Stokes is still translated, as Stokes suggested, as ‘Fogamur, it is a name for the final month in the autumn’ (although a wholly literal rendering would be more like ‘Autumn, that is a name for the last month in autumn’).³ That used by O’Donovan and Rhys, which seems older, reads ‘Fogamhar, i.e., it is for the final month that is used as a name’, with the immediate gloss ‘i.e., as if it were fogaimuir, under/within winter, i.e., the months of winter are long’ (Meyer 1994, 49). Their rendering of it is therefore entirely tenable, but the problem is what the author(s) meant. This was one of the points in the glossary at which the interest was to suggest meanings apparently embedded in the words being explained, and so the one on ‘fogamhar’ is then glossed with the meaning of ‘the time which is next to/leads to/is subject to winter’. So, the author(s) clearly thought of autumn as the last period of something, perhaps of the farming cycle, or simply (and conventionally) as the final season of the year, but it is hard to tell what. As a result, the foundational piece of data for the construct of a Celtic New Year is unsound.

The story of the bonfire at Tlachtga has a different problem: that it occurs only in one very late text, from the seventeenth century. This is the history of Ireland by the Catholic priest Geoffrey Keating, who stated that every Samhain Eve Druids would gather at Tlachtga from all over Ireland (it is near the centre of the land) to make a burnt sacrifice, and all the Irish had to rekindle their own fires anew from theirs (Keating 1857, 300). This tale enabled Keating to promote an image of an essentially and traditionally united Ireland, in the face of Protestant English conquest, but its provenance is unknown, and it does not feature in any earlier source. Its value as an authentic tradition is therefore very doubtful (see e.g. Binchy 1958, 129–30 for a dismissal of its worth), and it is not explicitly a New Year ceremony.

The evidence provided from modern folklore is likewise doubtful and ambiguous. In Rhys’s time the British Mummers’ Play was thought to be prehistoric in origin, and therefore its traditional seasons of performance could be thought to reflect those in antiquity. It is now thought to have been an innovation of the mid-eighteenth century, often spread by printed texts, which reached a peak of popularity in the early nineteenth century (Smith 1981; Fees 1984; Boyes 1985; Harrop 2021). This being so, it is likely that it reached the Isle of Man from mainland England shortly before Kelly encountered it. The season of its performance on the mainland was overwhelmingly at Christmas and New Year, and so the Manx mummers, while preferring to stage it at Hallowe’en, could have been using a text with that date in it. Later, the words would have been adjusted, as Rhys found. The early nineteenth-century farmer who told his hands that Hallowe’en (Sauin in Manx) had marked the New Year could have been making an idiosyncratic deduction of his own, based on the expiration of tenures and contracts then; or he might even have known Kelly. That custom of expiration would suit a season at the end of the annual farming cycle, when the harvest was gathered and the summer pastures finally closed. All this is pure speculation, but it does provide plausible alternative explanations to those of Kelly and Rhys. This is the more significant in view of the broader context: that none of the major collectors of Manx folklore during the eighteenth and nineteenth centuries, themselves resident in the island, found any tradition linking the opening of November to the New Year (Waldron 1865; Woods 1811; Train 1845; Moore 1891). One local folklorist, inspired by Rhys, made a systematic search of Man in 1896, seeking any trace of one, and discovered none (Sandle 2009).

Frazer’s use of folklore and comparative anthropology is likewise inconclusive. It is certainly true that many human societies, scattered across the world (but not all), have held seasonal feasts to honour their dead, and that many of those (at least in South-East Asia) have done so at New Year. It is therefore a good inference that the peoples of ancient northern Europe would have done at least the former of those things, but not a certain one. The number of modern European folk customs enacted around the start of November and related to the dead, which Tylor and Frazer noted, may possibly testify to the existence of an ancient pre-Christian festival at that time. It may, however, even more plausibly be a consequence of the historically undoubted major Christian festival of the dead, that of All Souls (often commencing on its eve, All Saints’ Day), which was—and in many regions still is—held then. The undoubted connection between Hallowe’en and divination, moreover, does not necessarily support the conclusions that Rhys and Frazer drew from it. For one thing, their assumption that divinatory rites are a major aspect of New Year festivals is not firmly grounded. The undoubted example of such a festival in historic Europe (and in ancient Rome) is that at the junction of December and January, which is the global one today, and while this one has some such rites, it has not been especially noted for them. Instead, the midwinter New Year was particularly fertile in rites of blessing, purification, and the bestowal of good fortune, such as burning of juniper in Highland Scotland, the first foot over the threshold in northern England and Lowland Scotland, and wassailing in southern England (Hutton 1996, 34–53). Hallowe’en was much less noted for these.

Nor were many of Hallowe’en’s divinatory rites peculiar to it. Those conducted by young people (especially women) seeking to know their future marital prospects were also found in Britain or Ireland on St Agnes’s Eve in January (immortalized by the poet John Keats), Midsummer Eve in June, and St Faith’s Eve in October, among other dates. The foretelling of who would die in the community in the coming year was also a feature of St Mark’s Eve in April, and occasionally of other times (Hole 1941; Roud 2006; Blackburn and Holford-Stevens 1999; Opie and Tatem 1989; Danaher 1972). There does, however, seem to have been an unusual concentration of rites at Hallowe’en designed to predict death (for which, to the titles earlier, may be added Brand 1908, 389–90; Ramsay 1888, 437), and a functional reason can be proposed for this. Hallowe’en opened the most lethal of all seasons in these climes, in which cold, damp, hunger, poverty, and disease would in normal years all be at their greatest, and mortality accordingly at its highest. People would accordingly choose that as a particularly relevant moment at which to attempt to discover who was going to survive, and who was not.

This leaves Frazer’s hypothesis concerning the origins of the Christian feasts of All Saints and All Souls. There was never any objective evidence for it: not a single medieval record of churchmen either attempting to suppress a pagan feast of the dead in northern Europe by instituting All Saints’ Day or attempting to replace one by instituting All Souls’ Day. Nevertheless, when Frazer proposed it, the intellectual climate was favourable, because towards the end of the nineteenth century a conviction had grown up among British scholars, based on a wide range of disparate pieces of evidence, that medieval Christianity had been merely an elite veneer covering a society in which most ordinary people remained at least partly pagan. Towards the end of the twentieth century, however, that conviction disappeared (for a recent summary of these successive developments, see Hutton 2022, 1–40). Most, if not all, scholars would now agree that medieval Christian cultures and societies had absorbed a great many features of the older religions, including beliefs, stories, images, and practices, but most, if not all, would equally reject the idea that paganism itself remained a powerful force at any level of society.

Instead, a different narrative has been established for the institution of All Souls, firmly based in written records and in Christian theology (for which, see especially Le Goff 1981; Brown 2015).⁴ It commences with a striking feature of that theology: that whereas pagans prayed to the dead, Christians, embracing a religion of salvation and damnation, prayed for them. By the third century, accordingly, the followers of Christ already believed in the efficacy of such prayers for assisting departed souls with their progress through, and status in, the afterlife. By that century the Christian polemicist Tertullian also laid great stress on offerings made on behalf of the dead upon the anniversaries of their demise, and suggested that ordinary Christian souls might reside in a place between heaven and hell. The greatest of late antique theologians, Augustine of Hippo, was convinced that the prayers of the living could influence the Christian God in deciding the fate of the dead. By the seventh century it was agreed that a fire of judgement purged the souls of some sinners, over a set period which could be reduced by prayer. Monastic communities recorded the names of individual dead members to be prayed for, on rolls, from the eighth and ninth centuries.

The Cluniac monastic network (founded in 910) was, however, still very innovatory in expanding the benefits of the liturgy to all the Christian dead, probably between 1024 and 1033. It attributed this decision, seemingly correctly, to the famous abbot Odilo, in office from 994 to 1049, and recorded a foundation legend for it. This recounted how a monk returning to France from Jerusalem had met a hermit living near a volcano, who told him that demons tortured the souls of sinners there. He had heard the devils lamenting that the prayers of monks delivered too many of their victims from torment, and that those of the Cluniac family were the most efficacious. He asked that the Cluniac monks extend their intercession to more beneficiaries, whereupon Odilo, on hearing the tale, instituted the feast of All Souls among them.⁵ Its date was probably chosen because of the proximity of the existing feast of All Saints, because saints were regarded as especially effective intercessors on behalf of the prayers of Christians. The huge prestige and influence of the Cluniac fraternity of houses ensured the dispersal of the new festival throughout most of Western Christendom over the next hundred years. During the following century, the concept of purgation of dead sinners, thus reinforced, was to flower into the full-blown official doctrine of Purgatory (Le Goff 1981: Wei 2012, 185–220). Instead of a popular pagan pressure on the Church to agree to such a festival, one recent scholar has suggested a popular Christian pressure for it, and for the developing theology that underpinned it: from ordinary clergy representing the wishes of their parishioners for a third location for souls to offset the harsh polarity of heaven and hell, and for an enhanced means by which the living could help ease the suffering of the deceased (Gurevich 1988, 114–18).

None of this is intended to denigrate Rhys and Frazer. Both were great scholars—audacious, dynamic, creative, and interdisciplinary—other aspects of whose work have so far withstood the tests of time. In the case of their theories about the Celtic New Year and feast of the dead, however, it seems necessary to conclude that the evidence they provided is certainly not adequate to provide conclusive support for these, and not even sufficient to provide much likelihood of their veracity. It remains to be seen, therefore, if any such likelihood can be established from sources that did not catch their attention.

Other Evidence

The first piece of data which could be taken to have relevance to the question is one of the orations of the last pagan Roman emperor, Julian, who reigned from 361 to 363, being his hymn to the sun god. At one point in this, he declares that the ‘beginning of the cycle of the year is placed at different times by different people. Some place it at the spring season, others at the height of summer, and many in the late autumn’ (Wright 1913, 155). This suggests that in ancient times many societies were celebrating their New Year at what might have been the junction of October and November, if Julian’s ‘autumn’ went that far, as his Greek term, ὀπώρα (opōra), could mean late summer; but Julian, alas, does not tell us which. Turning to medieval Irish and Welsh literature, not a single text refers to such a celebration, but there are a couple which might suggest one. The first is is a famous passage from the Irish tale Tochmarc Emire, ‘The Wooing of Emer’. In one part of it the eponymous heroine challenges her suitor, Cú Chulainn, to prove his worth by undertaking a task ó samsúan co hoimelc, ó oimelc co beltine, co bron trogain ó beltine. This may be rendered as from Samhain to Oilmelc (1 February), from Oimelc to Beltine (1 May), and to Bron Trogain (1 August) from Beltine (Van Hamel 1933, 31–32).⁶

Tochmarc Emire survives in two main recensions, the earlier and shorter of which lacks the passage concerned. The longer one which includes it is in Middle Irish, and probably eleventh- or twelfth-century (Meyer 1888, 1890; Van Hamel 1933; Carey 1989). This makes it less likely that this part of the story dates from a period at which memories of pagan practice would have been preserved. The cycle of dates cited in it looks like an annual one, starting at Samhain, but this is not made explicit. If treated quite literally it is actually nine months, not a full year, and it is in theory possible that Samhain was just the next festival after the challenge was issued. Alternatively, Samhain was certainly the opening of winter, and this could easily be treated as the first of the seasons, the one that prepares the land for those to come. Indeed, it would be the first in calendrical terms, for a culture which dated its years from midwinter. So it is not good evidence for a dating of years from Samhain. Moreover, the same story contains the passage ‘Samain, i.e., summer end, i.e., the end of summer is then. For it is from it that there is the division that used to be on the year back then, i.e., the summer from Beltine to Samain, and the winter from Samain to Beltine’ (Van Hamel 1933, 43; translated for me by John Carey). This is the classic dual division of the pastoral year, into the season when the pastures are open between May and October, and that in which they are closed between November and April. There is nothing in it, however, to suggest that the author was marking the beginning of a year from either.

The second text is equally famous, being Acallam na Senórach, ‘The Colloquy of the Old Men’, which is probably twelfth- or thirteenth-century.⁷ The relevant section concerns the great legendary queen of Connacht, Medb or Maeve, and tells how at Samhain she would go to a particular fort ‘to confer with her magicians and poets in order to learn that which during the coming year should turn out either well or ill for her’ (O’Grady 2000, 202). Again, this might be a New Year rite, or it might not. It certainly shows that the association between Samhain and divination goes back deep into the Middle Ages in Ireland, if not still further back in time. However, as indicated, divination alone is not a decisive indicator of the opening of a year; and this passage concludes the ancient and medieval texts possibly related to the question of a distinctive Celtic New Year that were missed by Rhys and Frazer. In medieval Ireland, as across the rest of Europe in historic times, the popular and traditional New Year started on 1 January, and the administrative and ecclesiastical one, when new officers took post, on 25 March or at Easter.

It may be asked, then, how Samhain was characterized in medieval Irish literature. The answer, which is well known to specialists, is that it was a time for major assemblies and feasts, as the year’s period of farming, herding, sailing, and fighting closed down to make ready for winter. It was also one when supernatural and superhuman beings were especially active, and inclined to interfere with humans. As such, it functioned as the most important, as well as the most supernaturally dangerous, festival of the pagan Irish year, as represented in the stories written by Christian authors who set out to remember or recreate the pre-Christian society of their land. It is especially prominent, as such, in tales such as Serglige Con Culaind, ‘The Wasting Sickness of Cú Chullain’; Mesca Ulaid, ‘The Intoxication of the Ulstermen’; Tochmarc Étaíne, ‘The Wooing of Étaín’; Macgnímartha Find, ‘The Boyhood Deeds of Finn’; and Echtra Nerai, ‘The Adventure of Nera’; but also in several others (for general considerations of the festival and its later development, see Hutton 1996, 360–85; Nagy 2003; Rogers 2002; Santino 1994). On the other hand, the equivalent festival in Wales, Calan Gaeaf, has no such importance in the medieval literature of that nation. Instead, its place as the point of the year at which supernatural and superhuman entities are most active and most inclined to enter the human realm is taken by the feast at the opposite place in the calendar, May Day or Calan Mai, and its eve, at the opening of summer. It has that function in such well-known prose tales as Pwyll Pendefig Dyfed, ‘Pwyll Prince of Dyfed’; Culhwch ac Olwen, ‘Culhwch and Olwen’; and Lludd ac Llefelys, ‘Lludd and Llefelys’. By contrast, the function of Samhain as the most frequently mentioned time for gatherings and feasts in early medieval Welsh literature is ‘the New Year’s feast’. No indication is usually given of its place in the calendar, but there are clues in the second and seventh stanzas of the poem traditionally known as Edmyg Dinbych, ‘The Praise of Tenby’ (for which the best current analysis is Russell 2017). In the former stanza it is at a time of stormy seas, which suggests winter, and in the latter it includes the making of a kelenic, a New Year’s Gift, being a custom celebrated across medieval Britain (and most of Europe) on 1 January (Hutton 1996, 12–13).

In nineteenth-century Welsh folklore, Hallowe’en, Nos Galan Gaeaf, was certainly a supernaturally dangerous place, with such horrid spiritual entities as a headless white lady and a tail-less black sow abroad in it. As such, it was the scariest Ysbrydnos of the year, a night on which spirits were abroad and gathered at places such as churchyards, stiles, and crossroads: the other two were May Eve and Midsummer Eve (Owen 1974, 131–41; Jones 1979, 146–51). It is tempting, and probably correct, to conclude that this reputation derived from its ancient and medieval significance; but this does not explain the apparent discrepancy with the medieval Welsh sources. It is also unwise confidently to back-project modern folklore onto the remote past as the Victorian and Edwardian scholars did, and perhaps we should not discount the impact on folk attitudes to the date of its long linkage to a Christian feast of the dead. Certainly, the nineteenth-century Welsh reputation of the night as infested by potentially menacing spirits matched the nineteenth-century Irish one of it, which of course directly continued medieval Irish tradition (Danaher 1972, 200–208). Moreover, it had exactly the same connotations in the same period in the Northern Isles of Scotland, where a Norse rather than a Celtic culture prevailed, and the ‘trows’ (trolls) were believed to emerge on that night and cause supernatural harm to farms (Nicolson 1981, 141–42). In this case, there is a medieval backdrop to such a belief, for the saga literature of Iceland represents the feast to bring in the winter season—the ‘Winter Nights’ which fell in mid to late October—as the greatest of the pagan Scandinavian year, the equivalent to the Irish Samhain. In some of this literature, also, such as Jómsvíkinga saga (chapter 2 in all editions), there is an association between the Winter Nights and divination. In that saga the earl of Holstein’s daughter Thyri is wooed by king Gorm of Denmark, and informs him that he must report to her the dreams that come to him during the three nights of the festival, and these will tell her if they are fated to wed (Blake 1962). It may well be, as suggested, that the imminence of the most frightening of the seasons freighted the festival that brought it in, across much of northern Europe, with a numinous significance. In none of the medieval literatures that purport to represent pagan times for their societies—Irish, Welsh, or Icelandic—on the other hand, does there seem to be any clear association between this festival and the human dead.

It is likely that the pre-Christian societies of ancient northern Europe, in common with so many across the world, had an annual festival to commemorate those dead, but no direct evidence for one survives. Instead, the ancient society in Western Europe which has left the clearest record of such calendar customs is the Roman, thanks mainly to Ovid’s Fasti. He portrayed two relevant festivals at Rome, held in the spring and early summer. The first was the Feralia, on 18–21 February, during which families honoured their own departed forebears and brought small gifts—a tile wreathed with garlands, a few grains of salt or grain, bread soaked in wine, or flowers—to tombs or the sites of funeral pyres. These were delivered with prayers and appropriate ceremonies. The regular temples were closed and the services to deities suspended, and if these dues to the dead were not offered, it was believed that many of the living would die within the year. Ghosts were thought to wander abroad during this time (Ovid 1996, 2.533–615). The second such feast was on 9–11 May, the Lemuria. Ovid held that this had originally been the time when all of the dead were honoured, but that the respect paid to the honoured departed was moved to February, while the older May rites were confined to getting rid of unwanted ghosts. A process of exorcism was staged annually for this, in which a household member laid a trail of black beans, which led to the door of the house, where metal vessels were clashed together to scare the phantoms out (Ovid 1996, 5.419–545).

It may be of interest now to see what can currently be traced of the development of the Christian feasts of All Saints and All Souls. Frazer’s account of it represented a simplification of the story given in the sources that he used, which were all secondary (Binterne 1829, 493–95; Smith and Cheetham 1875, 56–57; Herzog and Plitt 1877, 303). During the course of the twentieth century, matters have only got more doubtful and complicated, so that apparently the latest authority to sum up on the matter has declared that ‘how a feast of all the saints came to be celebrated on November 1 has not yet been demonstrated’ (Smith 2003, 289). It seems clear that by the seventh century there were regional Christian traditions of holding an annual feast to commemorate all Christian martyrs, which may have included saints in general. One of these was seemingly held at Rome on 13 May from the year 609 or 610 onwards (Thursten and Attwater 1956, 234–35). Another is mentioned in the Comes of Wurzburg, a German text of the late sixth or early seventh century, as taking place on the Sunday after Pentecost (Smith 2003, 290).

The problem with the appearance on 1 November in the work of Bede is that it does so only in manuscripts of that work compiled around a hundred years after Bede’s time, and the same difficulty attends its appearance in other texts that have origins before the year 800 but recensions only around or after that date (Binterne 1829, 493–95; Smith and Cheetham 1875, 56; Thursten and Attwater 1956, 234). The main text by Bede which features a commemoration of all saints on 1 November furthermore specifies that it was held in Rome:⁸ leaving aside the fact that Bede’s Anglo-Saxons could hardly be described as ‘Celtic’, this problematizes still further Frazer’s assertion that Bede’s work testifies to an early adoption of the November date in England. Furthermore, two closely-related Irish texts from the period 800–825, the Martyrology of Tallaght and the Martyrology of Óengus, record a feast of all European saints in Rome in April (under 20 April in all editions), but do not refer to one in Ireland (although see Hennig 1946, 1948 for an argument that this was a native tradition). Like other early Irish martyrologies, they start the year in January, although how much weight can be put on this as evidence for a native system is debatable, given that this genre of literature was based on Continental forerunners. Moreover, many entries in these works were added in subsequent periods, and the oldest manuscript of either is twelfth-century. As early medieval Ireland has been the greatest reference point for scholars seeking information on ‘the pagan Celts’, all this is also significant. The absence of such a festival, on the November date, from such a prominent eighth-century Latin Christian work as Chrodegang’s Rule, a variation on the Benedictine Rule written by a bishop of Metz, argues strongly at the least against any general knowledge of one in the Western Church at that time (Smith and Cheetham 1875, 56, where the text concerned is termed Chrodogany’s Calendar). It has been suggested that Pope Gregory III (731–41) consecrated a chapel to all the saints in the basilica of St Peter and fixed its anniversary on 1 November (Mershman 1907a). However, the source for this, in the book of biographies of Popes started in the ninth century, the Liber Pontificalis (chapter 92, section 6, in all editions), only says that he placed relics of the apostles, martyrs, and confessors in a chapel dedicated to the Christ and the Virgin Mary, and does not mention an anniversary (cf. Hennig 1948).

The first seemingly impeccable dating for an All Saints’ Day on 1 November seems to be provided by a letter from the leading scholar Alcuin of York, adviser to and biographer of the emperor Charlemagne, to Arno, archbishop of Salzburg in Austria, during 799, commending him for observing such a feast when presiding over a Bavarian council, although it is not clear whether Alcuin adopted it himself.⁹ He does not refer to any precedent for it. It was shortly after this that Pope Gregory IV (827–44) allegedly transferred the traditional Roman feast to commemorate the saints on 13 May to 1 November, on the grounds that it was easier to feed pilgrims visiting the city then, soon after harvest. This was at any rate asserted by the French liturgist and theologian John Beleth in the mid-twelfth century. It was Bishop Ado of Vienne, in the late ninth century, who recorded that Gregory asked Louis the Pious to impose the festival, at the new date, upon his whole empire for the sake of uniformity. Around 1100, the chronicler Sigebert Gemblacensis awarded the exact date 835 to this event. It was certainly a major innovation, being the first important Western Christian festival to be established outside the half of the year, between December and June, traditionally devoted to liturgical events. Aside from the purely local reason cited by John Beleth, there is no clear evidence as to why it was made, and what influence, if any, regional initiatives such as that by Arno contributed towards it. At any rate, its adoption by all branches of the Western Church was a slow business, extending to the end of the eleventh century, according to the scholar Sicard of Cremona, writing around the year 1200.¹⁰

The theological trajectory in the development of All Souls’ Day was considered earlier in this article, but it also has an institutional history which is only slightly clearer than that of All Saints’ Day. The early ninth-century German prelate, scholar, and politician Amalarius of Metz wrote that the liturgy for the saints was followed by that for the dead, but it is not clear whether he was speaking in terms of feast days, and if so, upon what dates (Smith and Cheetham 1875, 56–57; Thursten and Attwater 1956, 241). In the same period, Eigil, abbot of the major German monastery of Fulda, who died in 822, was later said to have instituted a service for all the Christian dead on 17 December (Candidus Bruun 1884, 99). The chronicler Widukind of Corvey mentioned a ceremony on 1 October to pray for the dead in Saxony in the late tenth century (Mershman 1907b). Odilo of Cluny therefore seems to have had various precedents for his more enduring annual commemoration in the early eleventh century. As in the case of All Saints’ Day, the pattern seems to have been one of individual local initiatives which eventually grew into a standardized practice, although the feast of All Souls was never formally adopted or ordained by a Pope in the manner of All Saints. In either case, there is nothing Celtic in the geography of this process. The early local initiatives seem to have been mostly Germanic, although it may be anachronistic to use such terms, and it could be better to think instead of a unified Frankish Christian culture covering the later France and Germany and associated with a dynamic new Holy Roman Empire.

Conclusion

In recent decades, historians and folklorists have come to appreciate the dynamism and creativity of culture, at all levels of society and through the centuries, and its propensity to draw for models on the past and justify innovations by reference to presumed past precedents, historically correct or not. Both have likewise come to agree that it is not their role to determine the legitimacy of present practices and ideas when evaluating the truth of historical claims made to support them. Thus, it is perfectly in order for people at the present day to celebrate a pagan New Year and/or commemoration of the dead at Hallowe’en, irrespective of the ancient precedents for one. It is only when those precedents are considered as a subject in themselves that a problem occurs, but the contention of this article is that a problem certainly exists. The argument for an ancient ‘pagan Celtic New Year’ and feast of the dead at that season was entirely the work of two very worthy and influential scholars writing over a hundred years ago, to which nothing of substance has been added since. On examination, the evidence for the New Year is all indirect, speculative, and inconclusive, even when items of it are included that escaped the notice of the original proposer of the concept. Evidence for the feast of the dead is pretty near non-existent. Neither idea has been conclusively disproved, and probably never will be: it is notoriously difficult to prove a negative in many areas of cultural history. Nevertheless, on the available evidence, it may be suggested that neither should find a place henceforth in publications on the Western European pre-Roman Iron Age, whether termed ‘Celtic’ or not, unless recognized as flimsy historical hypotheses, anchored in doubtful Victorian and Edwardian scholarship. Certainly, they should no longer be restated as proven fact.

Additional information

Notes on contributors

Ronald Hutton

Ronald Hutton is Professor of History at Bristol University, UK and Gresham Professor of Divinity at Gresham College, London. He is a Fellow of the Royal Historical Society, the Society of Antiquaries, the Learned Society of Wales, and the British Academy.

Notes

1 O’Donovan’s suggestion did attract some attention before Rhys gave his lecture; for example, being mentioned in the Gaelic Journal’s first issue in November 1882. I owe this reference, with thanks, to one of the anonymous readers of my article for the editor.

2 Although transcriptions of all manuscript versions are now available in the Early Irish Glossaries Database edited by Paul Russell, Sharon Arbuthnot, and Pádraic Moran at https://www.asnc.cam.ac.uk/irishglossaries/texts.php.

3 Stokes (1862, 20-22). I am very grateful to Dr Mark Williams for shaping my view of Sanas Cormaic and providing new and accurate translations of the relevant passages and for reading this whole article in draft.

4 I am very grateful to Dr Ian Wei for bringing me up to date on the scholarship in this area, and for reading this article in draft.

5 Jotsald, Vita Sancti Odilonis (Life of St Odilo), bk 2, ch. 13 (PL 1841, 142: cols 926–27); Peter Damian, Vita Sancti Odilonis (PL 1841, 144: cols 936–37).

6 Neither Rhys nor Frazer used this text in support of their argument for a Celtic New Year. O’Donovan did cite it, but not in support of that argument. Instead, he did so beforehand, as part of a discussion of whether it was possible to determine the season with which the pagan Irish year had commenced. He concluded, significantly, that it was not (thus far) (O’Donovan 1847, lii).

7 I am very grateful to Professor John Carey, for locating this reference for me after I had mislaid it, discussing current views on the dating with me, reading the whole article in draft, and providing the translation from Tochmarc Emire.

8 Bede, Martyrologia (PL 1841, 94: col. 1087).

9 Alcuin, Epistola (Letter) 91 (PL 1841, 100: col. 296).

10 John Beleth, Rationale divinorum officiorum (Rational for the divine offices), chap. 127 (PL 1841, 202: cols 133–34); Ado of Vienne, Martyrologium cum additamentis (Martyrology with additions), D. Kal. Novembris (1 Nov.) (PL 1841, 123: col. 387); Sigebert Gemblacensis, Chronica (Chronicles) R. et F. 21 (PL 1841, 160: col. 159); Sicard of Cremona. Mitrale (Liturgical commentary), bk 9, chap. 24 (PL 1841, 213: col. 414).