Intertextual Potential and the Reader in the Bevis Tradition: Boeve de Haumtone, Bevis of Hampton and Beuve de Hantonne

This article investigates the role that reader-oriented, unlimited intertextuality – that is, intertextual references only as perceived or even mis-perceived by readers – might have had to play in the transmission, adaptation, and development of the Bevis of Hampton narratives across English and Francophone traditions. In recent years, a number of studies have demonstrated the importance of the multi-text manuscript as a medium for texts to be integrated into new reading communities, and I therefore use multi-text Bevis manuscripts as a starting point for my enquiry. In particular, this article focuses on an Anglo-Norman Boeve de Haumtone manuscript (Leuven, Universiteitsbibliotheek, MS G. 170 or the ‘Firmin-Didot manuscript’, now lost), an English Bevis of Hampton manuscript (British Library, Egerton MS 2862) and a copy of the so-called “first version” of the continental French Beuve de Haumtone (Bibliothèque nationale de France MS Fr. 25516). Beginning with the “intratexts” that occur within the confines of the Firmin-Didot codex, this article moves forward to consider, in more hypothetical terms, the different textual encyclopedias that may have been available to the historic readers of Bevis manuscripts, and the ways in which readerly understandings (and misunderstandings) of a text’s intertextuality could prompt new narrative branches to sprout forth, and eventually lead to the emergence of new traditions.

In recent years, several studies have demonstrated that multi-text manuscripts enabled the translation, adaptation, and transmission of textual traditions across linguistic and geographical borders. Some researchers have taken a particular textual form as their focus, such as lyric or romance, while others have homed in on an individual narrative and its appearance in multi-text manuscripts across Europe (Pratt et el., 2017;Edlich-Muth, 2018;Zeldenrust, 2020). Popular texts surviving in an abundance of manuscripts are particularly valuable for tracing paths of transmission, offering the researcher a large body of physical data with which to work. The Bevis of Hampton tradition is just such an example: the original Anglo-Norman Boeve de Haumtone spawned a number of translations across a huge range of European vernaculars throughout the course of the Middle Ages and afterwards, from Middle English and Welsh to Yiddish and Byelorussian (Fellows 2017, I, pp. xv-xii).
If the multi-text manuscript was a format that could aid the process of textual transmission and adaptation across traditions, intertextuality evidently functioned as a highly effective method for integrating a text into existing literary traditions. The chanson de geste, described memorably by Dominique Boutet as a 'tentacular genre' [genre tentaculaire], exemplifies the power of intertextuality to draw texts into existing cycles and even to spawn new ones. 1 Thus, the continental French versions of Bevis of Hampton's narrative include personal and place names from existing cycles, such as Oliver and Roland, that are not present in the older Anglo-Norman Boeve. 2 Eventually, Bevis would even be incorporated into new cycles such as the Franco-Italian Geste Francor, which followed the genealogy of Charlemagne back to Pépin. 3 The composers of the Middle English Bevis also used intertextuality for the purposes of integration. References to insular romance heroes, and even St George in the Auchinleck codex, were introduced to the Middle English Bevis in order to bring it in line with the genre expectations of romance in English (Weiss, 1979;Rouse, 2008, pp. 115 − 16;Barry, 2016, p. 299). 4 In the Bevis tradition, a problem that has not yet been solved is how to approach the question of intertextuality with regards to the earliest surviving version of the Bevis narrative, the Anglo-Norman version. This rendition of Bevis' story, unlike its Middle English and continental counterparts, exhibits no obvious intertextual refer-1 'A l'image critique usuelle d'un genre "contaminé" par l'influence d'autres poétiques, et qui dégénérait progressivement sous cette influence dissolvente, il convient donc de substituer l'image d'un genre tentaculaire, qui se nourrit et s'étoffe de ces annexions dynamiques' [The usual critical image of a genre "contaminated" by the influence of other poetics, and gradually deteriorated under this eroding influence, should therefore by replaced by the image of a tentacular genre, which nourishes itself and grows from these dynamic annexations.] Boutet (1993), p. 227. Boutet points to the Doon de Mayence tradition as a 'crossroads between epic cycles' [carrefour des cycles épiques]; see Boutet (1999), I, pp. 101-109; Cazanave (2014), 217-238. 2 For example, in the first continental French version, when Beuve is sold as a slave, the merchants advertise their wares by proclaiming him ' […] Mascitelli (2017). 4 Guy of Warwick seems to have been a figure particularly emblematic of "romance" in a Middle English context; see Liu (2006), p. 343. ences. Moreover, only three manuscript witnesses of Boeve survive: two of these are binding fragments; the third contains Boeve as its only medieval text. 5 This makes it difficult to assess how Boeve resonated intertextually "across" the multi-text manuscripts in which it was contained.
This article approaches this problem with regards to a multi-text Boeve manuscript that does not survive: the so-called Firmin-Didot manuscript, once held at the University of Leuven but lost to fire in 1940 (Olim Leuven, Leuven Universiteitsbibliotheek, MS G. 170). Though the unavailability of this codex poses a serious obstacle to further scholarly analysis, it was carefully studied and largely transcribed prior to its destruction. These studies furnish the most important evidence we have of what multi-text manuscripts containing the Anglo-Norman Boeve might have looked like, and are valuable for addressing the intertextuality of the earliest surviving version of the Bevis narrative. I suggest that potential intertextuality, even misunderstood intertextuality as a reader may have perceived it, could play an important role in a narrative's transmission -particularly when combined with the internal intertextuality particular to multi-text manuscripts. Using the Firmin-Didot codex as my starting point, and beginning with what we do know about the intertexts it once contained based on surviving studies and descriptions, I will then move onto more hypothetical considerations of what the lost intertextual knowledge frameworks (or "encyclopedias") of the Firmin-Didot's later readers might have looked like. By considering the Firmin-Didot in relation to other English and Francophone Bevis manuscripts, we can productively approach intertextuality in fragmentary traditions, and even lost manuscripts, by thinking about their intertextual potential. In centring both a recentlylost manuscript (the Firmin-Didot) and irrecoverable hyparchetypes as evidence, this essay resists what Neil Cartlidge (2005) has termed a 'rhetorical materialism' in medieval manuscript studies, 'according to which the demands of manuscripts for the attention of literary scholars is asserted in terms of their incontestable tangibility in the present' (p. 31).

Approaching Intertextuality
The study of intertextuality in fragmentary traditions, and within the context of lost manuscripts, requires some thought to be given to methodology. Colleagues engaged in the study of ancient texts have been grappling with these problems for some time. The editors of Between Text and Text: A Hermeneutics of Intertextuality in Ancient cultures summarise the issue: 5 The binding fragments are Glasgow University MS Hunter 466 (olim V.6.16) and Lambeth Palace MS 1237 nos. 1-2. The other codex is Bibliothèque Nationale de France, MS Nouv. Acq. Fr. 4532. I do not discuss this latter manuscript here because this essay concerns intertextuality as it may have been interpreted by medieval readers, but this interesting manuscript raises questions about how Boeve, a text often called an "ancestral romance", might have been perceived as such by the person responsible for heavily annotating Boeve and binding it with a set of early modern English manorial records, a very different kind of text yet one still concerned with the theme of heritability.
the study of ancient intertextuality has [...] particular difficulties to address. To name just a few: the bulk of the ancient literatures is known only in fragments or is not preserved and our knowledge of the ancient cultures and languages is far more limited than the one of modern cultures and languages. (Bauks et al., 2013, p. 11) The same problem exists when we try to uncover intertexts in many Anglo-Norman romances: if we approach these texts to seek information about their direct sources, or to uncover intertextual allusions intended by their author(s), adaptors, and copyists, we will often find ourselves disappointed.
Biblical scholar Stefan Alkier (2013) addresses this problem by considering various possible approaches to the study of intertextuality (p. 295). The first, a "limited", "production-oriented" approach, considers 'only those textual relations which are written into a given text, or at least can be postulated on the basis of the signs collected in the text' (ibid). This requires several events to take place: a textual producer must consciously refer to another text; and this must be communicated successfully to their reader, including the modern scholar, since we are required to identify the intertext. In this definition there is no space for readerly interpretation; an intertext is only an intertext if an author has placed it there, and it is the job of the scholar to show that they did so intentionally.
This limited, production-oriented approach is not ideal for investigating intertextuality within a manuscript context, rather than an individual textual tradition. When considering intertextuality as a tool for transmission, miscommunication must be accounted for, and this is not a consideration in a production-oriented view of intertextuality. Miscommunication could have happened at any stage of a manuscript's production or reception. A scribe might have misunderstood a personal or place name and attempted to correct it, leading to readerly misunderstanding. 6 Alternatively, a reader might simply have misinterpreted a text, perhaps perceiving allusion within it that was not really there (at least in a production-oriented sense) because it was never intended by the textual producer. Miscommunication became considerably more likely as a manuscript descended through time and across geographical space, being interpreted by new readers in completely different temporal, linguistic, geographical, and cultural contexts to those in which it was originally produced. Medieval manuscripts had long afterlives, often progressing through a series of owners, some of whom left traces while others did not. 7 A fourteenth-or fifteenth-century reader would have had a very different frame of reference ("encyclopedia") to a contemporary scholar, or even a twelfth-century author. 8 Particularly for texts with such a slippery sense of history as the Bevis narratives, it may have been difficult for a reader to contextualise them in time, creating the perfect circumstance for anachronistic misunderstanding; thus, a reader's mistaking Emperor Doun of Germany (Anglo-Norman Boeve) for another Germanic Doon (Doon de Mayence, continental Beuve) could cause a new narrative branch to sprout, as I will go on to suggest. 9 Such misunderstandings could be powerful events for generating new meaning, and therefore new traditions.
Fortunately, there are other ways of understanding intertextuality. In contrast to a "limited" approach, an "unlimited" methodology is divorced from authorial intent and aligns instead with the position, rooted in Kristevan dialogics, 10 that a given text stands in a relationship with the entire universe of texts, including those which were produced after it and even those which are still to be produced. A single text is not an autonomous entity, but rather is integrated into an endless, unpredictable, and therefore indomitable multitude of interwoven connections with other texts, which are constantly shifting. This makes its meaning uncontrollable. (Alkier, 2013, p. 295, emphasis mine) Beyond the limited/unlimited binary, Stefan Alkier argues that we might consider intertextuality in terms of whether it is "production-oriented" (that is, placed there intentionally by an author, adaptor, or copyist), "reader-oriented" (intertexts as perceived by a reader), or "experimental" (akin to Kristeva's "only the text" model). A reader-oriented perspective can itself be either "limited" or "unlimited": 'you can ask for intertextual relations that concrete readers have made [for example, exegesis] or imagined readers could have made' (Alkier, 2013, p. 301). An unlimited readeroriented perspective therefore considers intertexts through the eyes of readers rather than producers, asking only for intertextual relations that imagined readers could have made. Approaching the intertexts of the lost Firmin-Didot manuscript in this way (and in reference to other Bevis manuscripts) allows us to reconstruct some of the intertextual connections that might have occurred to its historic readers, prompted by a familiar character or place name, a cultural reference, or an episode lifted from another literary work. This approach also allows us to take into account miscommunication by considering the codex's intertextual potential "out of time", detached from the context of its original production. By paying attention to Firmin-Didot's possibilities for intertextual interpretation, we can better understand the ways in which the development, translation, and adaptation of the Bevis tradition across medieval Europe may have been shaped by potential intertextualities.

The Firmin-Didot Manuscript and its Intertextual Potential
The Firmin-Didot manuscript probably dated to c. 1250-1300 (Martin, 2014, p. 10;Le Person, 2003, p. 26). A single-column manuscript of 'medium quality' at best, it contained the Anglo-Norman Boeve followed by a copy of Fierabras, a Matter of France text either contemporaneous with Boeve or slightly earlier in date (Weiss, 2008a, p. 159). Though the Firmin-Didot Fierabras was never transcribed in its entirety, it may have been quite close to the version preserved in Bibliothèque de l'Escorial M. III-21 (Hardman & Ailes, 2017, p. 268). Firmin-Didot was acephalous, but it may well have contained only these two texts; Phillipa Hardman and Marianne Ailes (2017) suggest that manuscripts containing chanson de geste "pairs" were more popular in England than the cycle manuscripts preferred on the continent (p. 149). Although Firmin-Didot does not survive, it remains the most useful Anglo-Norman Boeve codex for comparison with English and continental multi-text Bevis manuscripts.
Both texts in the manuscript share a lot in common: firstly, they have both been identified as insular chansons de geste or romance/chanson de geste hybrids, exhibiting features that distinguish them from continental chansons de geste. 11 Both texts were likewise translated into Middle English romances-Boeve became Bevis of Hampton, while Fierabras exists in three Middle English translations (Hardman & Ailes, 2017, pp. 265 − 66). Both texts feature Christian heroes who convert "Saracen" opponents to Christianity; both feature a Bele Sarasine love interest; and both involve a narrative of recovery-Bevis of his stolen inheritance; Oliver and Charlemagne of a set of holy relics taken from Rome. Each text has a pagan giant as a main character who is later converted (Ascopard and the eponymous Fierabras respectively). The Anglo-Norman Fierabras contained in Egerton MS 3028 describes the French fighting a people called the 'Ascoparz' at a point in the text not recorded in the Firmin-Didot Fierabras, so there may have been more specific resonances between the two texts. 12 The action of both takes place, for the most part, around the Mediterranean basin, at the interface of the Christian and Islamic worlds. What all this suggests is that, in addition to any observations we can make about references made in Firmin-Didot to texts beyond the manuscript, its two internal texts also share much in common.
While we cannot draw any certain conclusions about how Firmin-Didot was read intertextually, we can make some suggestions. Fierabras is a Matter of France text relating specifically to Charlemagne and his peers: Charlemagne, Roland, and Oliver all appear as main characters, and the events of the Chanson de Roland are directly referenced at the end of Firmin-Didot's Fierabras. 13 We can therefore consider other 11 K. V. Sinclair and Marianne Ailes respectively have argued that the version contained in Firmin-Didot is a distinctly Anglo-Norman production with several formal and narrative differences. Sinclair (1993); Ailes, (2006). Judith Weiss, Marianne Ailes and Jean-Pierre Martin have also argued that the Anglo-Norman Boeve exhibits the formal, generic and stylistic features of a chanson de geste adapted for insular tastes, or else a romance/chanson de geste hybrid; see Weiss (2008a); Ailes (2008); Martin (2003). 12 The 'Ascopart' people also appear as a pagan people group in a large number of other chansons de geste; see Langlois (1904), p. 5. 13 See the final 25 lines transcribed in Stimming (1899), pp. v-vi. texts within the geste du roi cycle as part of Firmin-Didot's intertextual world. 14 The Anglo-Norman Boeve does not appear to belong to this cycle, but if it were frequently copied alongside Charlemagne texts, as it is in Firmin-Didot, there is no reason why this might not have led to it being associated with Matter of France cycles as a result. Charlemagne-related texts are copied alongside Bevis texts in not only the Anglo-Norman tradition (Firmin-Didot) but also in manuscripts containing Bevis' Middle English, continental French, Welsh, Irish, and Franco-Italian versions, several of whose Bevis texts were translated from the Anglo-Norman. 15 A text that may have been directly based on an early prototype of the Anglo-Norman Boeve, Daurel et Beton, is set firmly in the world of Charlemagne (Martin, 2014, pp. 20-21). 16 Naturally, continental versions of the Bevis narrative go even further in adapting Bevis into the textual world of the continental gestes; but this is not to say that this process might not have had its roots in Anglo-Norman England. However, the paucity of surviving Anglo-Norman Boeve codices, and the fragmentary state of what does survive, prevent us from investigating this matter any further.
The Firmin-Didot Boeve was, unlike Fierabras, edited in its entirety. It appears to contain no explicit intertextual references, but it may simply be that we cannot perceive them because its producers' frames of reference ("encyclopedias") are not fully recoverable, and it is these frames of reference that encode intertextuality for communication from producer to reader. There are also several moments in the Firmin-Didot Boeve (and, indeed, other Bevis manuscripts) where it is difficult to discern whether a reference is intertextual or more generalised cultural allusion. These incidences demonstrate how futile a limited production-oriented approach would be for studying intertextuality in Boeve. Let us consider the "Dry Tree", a seemingly meaningful location that Bevis claims to have visited in the Anglo-Norman Boeve. 17 The "Dry Tree" might represent a production-oriented intertext; but it may simply be an allusion to a very widespread cultural tradition, as scholars such as Rosanne Gasse (2013) and Rose Jeffries Peebles (1923) have persuasively argued. The travels 14 The Geste du roi category is memorably described in the prologue of Bertrand de Bar-sur-Aube's Girart de Vienne, alongside other categories of chanson such as the Geste de Doon de Mayence; see Van Emden (1977), laisses 1-3, cited in Boutet (1999) of Marco Polo suggest that both possibilities exist, stating that the "Dry Tree" is a place named in the "Book of Alexander" but also that "the Christians call [it] the Dry Tree" (Gasse, 2013, p. 71). The thirteenth-century romance King Alisaunder features a "Dry Tree", and this may have been borrowed from an earlier source known to the Boeve author. 18 Ultimately, without a clearer knowledge of the unrecoverable encyclopedias of the text's twelfth-century producer, we cannot know whether the "Dry Tree" reference represents an intentional intertext or simply a cultural allusion.
Another seemingly specific tree, "The Great Tree" [l'arbre grant], appears elsewhere in Boeve. It is located at Saint Gilles, the pilgrimage site where Boeve's ally Sabaoth is reunited with Josiane. 19 Unlike the "Dry Tree", this "Great Tree" is a complete unknown: no scholar is sure of what this "great tree" of Saint Gilles might be, and it is never explained in the text-perhaps a reader would simply be expected to recognise the reference. Saint Gilles itself is meaningful as a setting, however, when considered from an unlimited, reader-oriented, and diachronic perspective. It is the birthplace of Élie de Saint Gilles, an epic hero with his very own chanson de geste pair or mini-cycle sometimes known as the Geste de Saint Gille, composed in the late twelfth century around the same time as Boeve. 20 This pairing includes Élie de Saint Gilles and Aiol, and is preserved in a single manuscript, Bibliothèque Nationale de France MS Français 25516, where it appears alongside a continental French version of the Bevis narrative, Beuve de Hantonne, and the romance Robert le Diable.
The question of whether the Geste de Saint Gille would have been known to the Firmin-Didot's readers is impossible to answer. We can at least say that the continental readers of the Geste de Saint Gille knew of the Bevis narrative, since the continental Beuve de Hantonne appears alongside the Geste in MS 25516. In fact, there is some evidence that one of MS 25516's owners, in the early fifteenth century, intentionally changed the shape of the manuscript to bring Beuve and the Geste de Saint Gille together. The history and provenance of this luxurious manuscript is preserved in great detail. Copied c. 1250-1300 in Northern France, it was owned by, and most likely produced for, the aristocratic court of Flanders. The original patron of the codex is not known, but a possible candidate is Gui de Dampierre (c. 1226-1305) or his immediate family members. Literary activity flourished at the Flandrian court during Gui's time, and the family indulged in a luxurious lifestyle, appearing to have been particularly interested in rich objects, literary patronage, and musical entertainment (Stanger, 1957, pp. 222 − 23). The manuscript's cohesive "stalk border" page design and rich illustrative scheme create the impression of a cohesive, carefully planned manuscript that has survived exactly as it was initially designed (Malicote, 2010). However, the manuscript's collation suggests it was intended to be detachable, with its constituent texts occupying their own distinct booklets. Each text begins at the start of a quire and ends on quire-end leaves, and text-end quires (apart from the division between Elie and Aiol) contain an odd number of leaves, suggesting that leaves have either been added or cut away. 21 This codex would therefore have been far easier to customise than fixed-order Bevis manuscripts, such as the Middle English romance collection Egerton 2862 (Bateman et al., 2022). The detailed inventories of the Dukes of Burgundy appear to suggest that 25516 did indeed change shape. It was in its present form by 1420, when it was included in an inventory made on the death of Jean sans Peur; but a 1405 inventory seems to list the codex as 'ung livre de Sebile, d'Ayeul et de Helie' (Hughes, 1978, 182 − 83;Derolez et al., 2016, p. 115, p. 141). None of the texts in the manuscript as it survives might feasibly described as a "livre de Sebile"; this may have been the Chanson de Sebile that only survives in fragments, or perhaps a lost text on which both the Chanson de Sebile and the fourteenth-century Macaire are thought to have been based (Resoort, 1998, pp. 248 − 49). Macaire appears in the Aiol mini-cycle as a treacherous antagonist, and is even illustrated at the end of 25516 at fol. 172v. Of course, the earlier inventory might simply have omitted Beuve or Robert, and there is also a possibility that it refers to a different manuscript altogether-it was not unusual, in the fifteenth century, for the book collections of the Burgundian nobility to include more than one copy of a text (Wijsman, 2013, p. 85). 22 However, the detachability of this manuscript suggests the possibility that its subsequent owners may have decided to re-unite Bevis and the Geste de Saint Gille within its binding. At the very least, it is possible that the codex once contained other texts, such as the mysterious 'Livre de Sebile', that have since been detached.
Ne of Arthor ne of Gawayn Ne of Launcelet de lake Ne of Beues ne Gy ne Sydrake Ne of Ely ne of Octauyan [...] 23 (Egerton MS 2862, fol. 39r) [I will read you no romance of Partenope nor Ypomedon, nor of Arthur or Gawain, nor of Lancelot de Lake, nor Bevis, Guy, nor "Sydrake" or "Ely", nor Octavian.] Could this have been intended as a reference to Élie of Saint-Gilles? If it was, the name was not a familiar one to the scribes responsible for copying the other extant Richard witnesses, suggesting that if Élie ever had been widely known in England he did not remain so by the mid-to late-fifteenth century (or not, at least, to the readership of Richard). 24 Unsurprisingly, lesser-known names in the list cited above are those which vary from witness to witness: the copyist of Arundel 58 renders the unfamiliar Partenope as 'Perse ne of Pene', for example (Schellekens, 1989, I, A.4698). The other names to vary dramatically are are 'Ely' and 'Sydrake'. 25 The other witnesses of this list tweak these unfamiliar names, seemingly to fit minor characters in the Middle English Arthurian tradition. 26 'Ely' becomes 'Ury' in Gonville & Caius MS 175/96 and may refer to Sir Urry of Thomas Malory's Morte Darthur, though the Anglicana hand of the Gonville copy of Richard looks rather earlier in date (c. 1425) than Malory's Morte (c. 1469-1470). 27 The same manuscript renders 'Sydrake' as 'Sere Vrrake', while the London Thornton manuscript instead has 'Errake', a name Thornton uses across his manuscripts to denote the Arthurian knight Sir Erec. 28 The inclusion of 'Ely', 'Pertenep', and 'Sydrake' in the Egerton list is striking, not only because the philosopher Sydrac (of Le Livre du Sydrac) seems like a strange choice for such a hero list, as R. S. Loomis has pointed out, but because these figures are protagonists in Francophone literature and are not represented in any literature 23 I have capitalised names here, but otherwise this is a diplomatic transcription. The modern English translation is mine; names in double quote marks represent figures whose identities are debated, so I replicate them simply as they are spelled in the manuscript. 24 As this part of the text is missing in the Auchinleck manuscript, Egerton represents the earliest manuscript with the list containing 'Ely' and its variants.  Larkin (2015); and for Arundel MS 58 see Schellekens (1989). 26 An exception is the London Thornton manuscript's rendering of 'Ely' and 'Vly'. María Cristina Figueredo has suggested that this might refer to Ulysses (2009, p. 137, p. 192). 27 This is based on my own inspection of the manuscript; the relevant passage appears at fol. 47v (page 92 according to the old pagination). On the dating of the Morte D'Arthur see Crofts and Whetter (2019), p. 54. 28 'Errake' is named elsewhere in the London Thornton codex in the Parlement of Thre Ages (as 'Sir Errake Fytz Lake'), and also appears twice in the Lincoln Cathedral Library MS 91: in Awyntyrs off Arthure and the Alliterative Morte Arthure. See Moll (2003), p. 99, pp. 141 − 42. in English known to have existed at the time Egerton was copied. 29 But although the Middle English adaptations of Sydrac and Partenope's narratives postdate Egerton, we know that their earlier French versions were circulating in England. The Frenchlanguage Partonopeus de Blois survives in at least two Anglo-Norman codices that predate Egerton. 30 The French Livre de Sydrac even appears in a trilingual insular collection that also contains, within its folios, a reference to King Edgar as the killer of Bevis' horse. 31 It is not clear whether this reference was made with the Anglo-Norman Boeve or Middle English Bevis in mind. What is clear is that these threads hint at the lost texts that made up the multilingual fabric of thirteenth-and fourteenthcentury England, and the Geste de Gilles may also have been represented in this rich tapestry. Although no version of Élie's narrative survives in Middle English or Anglo-Norman, it had been adapted into Old Norse by the thirteenth century, perhaps from an Anglo-Norman original according to H. G. Leach; and Aiol's story was translated into Flemish, Middle Dutch, Italian, and Spanish between the 1200 and 1500. 32 It is not impossible, then, that the Saint-Gilles material might have been included in the encyclopedias of the Firmin-Didot's readers. Paying attention to Bevis manuscripts in terms of their intertextual potential calls attention to unknowable dark spaces and possible connection points where texts have been lost and past readers' encyclopedias might once have been found. 33 There is one final aspect of the Firmin-Didot's intertextual potential that requires comment: a character, Doun 'Emperor of Germany' [le emperor de Alemaine] (l. 26). Doun is Boeve's usurper father-in-law, invited over from Germany by Boeve's mother to murder Boeve and his father. In every continental version of Beuve, each adapted from the Anglo-Norman, the character has a fuller name: Doon de Mayence. 34 This Doon gives his name to an actual geste cycle known as the Geste de Mayence (or the Cycle de Doon de Mayence), a cycle that appears to have developed gradually over time as two distinct families of traitors and rebels merged, and which ultimately treats the subject of Doon and his twelve sons (Ailes, 1998, pp. 41-43;Negri, 2014, pp. 97-111). Doon himself was probably one of the last characters of 29 Loomis (1916), pp. 455 − 66. Despite Loomis' scepticism about Sydrac's inclusion in the romance hero list, Le Livre du Sydrac has sometimes been to as the "Roman de Sydrac". See Nichols Jr (1968), pp. 167 − 68;Holler (1986), p. 184. 30 Partonopeus de Blois survives in two Anglo-Norman codices: Norman collections, Yale University, Beinecke MS 395 fols. 129v-130v and Biblioteca Apostolica Vaticana, MS 1971, fols. 1r-59v. 31 Cambridge University Library, MS Gg.1.1. Le Livre du Sydrac appears at fols. 495r-627v. The reference to Edgar and Bevis appears in an Anglo-Norman Short Metrical Chronicle (early s. xiv) at fol. 488r, and is cited in Fellows (1986), p. 140. 32 Malicote (2006Leach (1921), pp. 152 − 53, pp. 259 − 60. 33 The sad state of surviving insular narratives and their representative manuscripts needs little introduction. The most up-to-date study, employing a model borrowed from ecology, suggests that English manuscripts have fared particularly poorly with only around 4.9% surviving. See Kestemont et al. (2022), pp. 765 − 69. 34 In version 1, preserved in MS 25,516 and edited by Stimming (1911), l. 14 (Do de Maienche). In version 2 (Stimming, 1912-18), ll. 159 − 60 ('Do de Maience'), ll. 243 − 44 ('Doon, / cel de Maience') and elsewhere. In version 3 (Stimming, 1914-20), l. 505 ('Dos de Maience'), l. 568 ('Do de Maience') and elsewhere. On the date of the three French versions and their relationship to each other and to the Anglo-Norman Boeve see Stimming (1911), pp. xi-xii and p. lii; Stimming (1914-20), II, p. 165. the cycle to be given a fully-developed story of his own; indeed, the chanson de geste titled Doon de Mayence was actually composed slightly later than Boeve. 35 Nevertheless, the Firmin-Didot's readers and owners may well have been familiar with the figure of Doon de Mayence, for copies of Mayence cycle texts survive that were circulating in medieval England (Ailes, 2014). If Boeve's insular readers did not make the connection between Emperor Doun and Doon de Mayence, the composers of the continental French Bevis texts certainly did. In efforts that call to mind Roland Barthes' 'The Death of the Author', the Chanson de Doon de Mayence poet, seemingly concerned that their protagonist not be confused with Boeve's nemesis, states twice with some alarm that this Doon de Mayence is not the same as Bevis' evil stepfather. In so doing, the author contradicts themself. First, Doon tells us in his own words that there are two places called Mayence:  Pinvidic (1997) suggests the second half of the thirteenth century as a date of composition (p. 207). 36 I extend my gratitude to Corin Corley and also to the anonymous review of this article for their helpful suggestions in the translation of these passages.
Ains est li anchïen et li premerain vis Dont la geste sailli des barons de haut pris [...] Pour chen le vous ait dit que n'en soie repris, Que tel i a de vous qui en estoit pensis (Pey, 1859, ll. 6650-6667) [Sirs, you know well, and I am completely sure of it, That there were many "Charleses" in Paris [in the past], In the great Narbonne there were many Aimeris, And in Orange similarly there were many Counts [named] Guillaume And, likewise, there were once many Doons in Mayence. This Doon, of whom I sing to you, who undertook these deeds Against King Charles [Charlemagne], and who defied him, Was not that same Doon, the false traitor Who chased Bevis of Hampton from his land, The husband of Josianne, the well-made and fair of face. Rather, it is the earlier one, the one who lived first, Whose geste emerged from barons of high nobility [...] I tell you this so that you don't reproach me, Because there are some of you who were wondering about it.] Although these protestations have been taken at face value by some scholars-such as the text's editor Alexandre Pey (1859, p. xv) -they should give us pause. The two passages appear to contradict each other: the first claims that the Doon of this text hails from a different Mayence, the one beside the Rhine (that is, Mainz), rather than the one close to Hampton across the sea. The second passage, however, suggests that both Doons were separate figures from the same location, compared to the many Aimeris of Narbonne or Guillaumes of Orange. Perhaps these interventions are the work of two different poets. Either way, they show a real concern that the text's protagonist should be mistaken for the villain of the Bevis narratives. This indicates that there was real misunderstanding on the part of Doon de Mayence's readers about these two homonymous figures, or that the poet expected such misunderstanding. And if texts from the Mayence cycle were indeed circulating in England, as Marianne Ailes has suggested they did, we must work with the possibility that this readerly interpretation-or misinterpretation-may also have affected readings of the Firmin-Didot.
In conclusion, while it may not be possible to identify any production-oriented intertexts in the Firmin-Didot codex, or even intertexts that occurred to any concrete readers, there is clearly value in considering this lost manuscript in connection with other manuscripts and texts in terms of its intertextual possibility. By being attentive to the afterlives of medieval manuscripts, it is possible to shed light on opportunities for intertextual interpretation and misinterpretation, and therefore on routes of narrative transmission and development. This means dealing with the uncertain, the slippery, and the anachronistic. But these nexus points, however invisible or unrecoverable, are precisely the points at which new meaning, and new traditions, were generated. The low survival rates of medieval manuscripts, particularly manuscripts containing epic and romance, mean that the chance of two cognate manuscripts being identified are slim, and the chances of cognates sitting across the divide between a source text and its translation even slimmer. This forces us to consider other approaches to the study of textual transmission. More experimental models, such as a reader-oriented unlimited approach to intertextuality, can help us to uncover how narratives such as the Bevis narrative may have worked their way across Europe thanks to the generative power of intertextual interpretation and misinterpretation.