Chapter 2
Bodleian MS Digby 23
In 1835 a young philologist named Francisque Michel was commissioned by the minister of public instruction, François Guizot, to visit England and transcribe a number of ancient works, including a poem on the Battle of Roncevaux that was known to exist in the Bodleian Library. Two years later Michel published La Chanson de Roland ou de Roncevaux du Xlle siècle publiée pour la premiére fois d’aprés le manuscrit de la Bibliothèque Bodléienne à Oxford, and the poem entered French literary history.1 Henceforth the standard version of the poem would be the text of the Oxford manuscript, and it would bear the title Michel had given it. The Song of Roland had been born.
Of course, the history of Roland’s death had never been entirely forgotten, especially in France. It was known through Dante, Ariosto, and Cervantes and through a popular tradition still potent enough for Claude Joseph Rouget de Lisle, author of the “Marseillaise,” to draw on it for another revolutionary song, “Roland à Ronceveaux,” which had the defiant refrain “Mourons pour la patrie.”2 Scholars knew of the account of the Pseudo-Turpin, the twelfth-century Latin chronicle allegedly composed by Charlemagne’s heroic archbishop, and of numerous other medieval references to the exploits of Roland, Oliver, and Charlemagne, and they knew that these exploits had been the subject of chansons de geste, including one version allegedly sung at the Battle of Hastings by a member of Duke William’s household, sometimes identified by the name Taillefer.3 The best-known account was that offered by the Anglo-Norman chronicler Wace in his Roman de Rou, composed sometime between 1160 and 1174. Wace describes Taillefer leading the Normans into Battle with his song:
Taillefer, qui mult bien chantout,
sor un cheval qui tost alout,
devant le duc alout chantant
de Karlemaigne e de Rollant,
e d’Oliver e des vassals
qui morurent en Rencevals.4
Taillefer, who sang very well, was mounted on a horse that raced along, and he went in front of the duke singing of Charlemagne, Roland, and Oliver and the knights who died at Roncevaux.
Wace’s story was noted by Claude Fauchet in his influential Recueil de l’origine de la langue et poésie françoise of 1581, by Voltaire, and by the British antiquarians Thomas Percy and Joseph Ritson, for whom it evoked the power and fascination of a lost oral tradition.5 Voltaire is perhaps the first to refer to Taillefer’s performance as a single song: “The old histories tell us that in the front rank of the Norman army, a squire named Taillefer, mounted on an armoured horse, sang the song of Roland, that has been for so long in the mouths of the French without the slightest trace remaining”6 This lost work was what so many early scholars hoped to find, not just another poem about Charlemagne and his twelve peers, but the very song of Taillefer.
By Michel’s day, the search for this work had been going on for some time. In 1777 the Marquis de Paulmy, chief editor of the Bibliothèque universelle des romans, a popular series devoted to summaries of “romances,” published an account of the stories of Charlemagne and Roland based on the Pseudo-Turpin. In his account Paulmy speculated that French troops going into battle might have sung the lost Chanson de Roland, and he actually went so far as to offer a possible reconstruction. As Paulmy explained, the Chanson de Roland could scarcely deal with all of Roland’s great deeds, and it therefore chose to present him as a model to imitate.7 The poem continues for eleven stanzas, praising Roland as a paragon: brave, modest, obedient, a good Christian, a moderate drinker, reluctant to seek a quarrel but a terror to his enemies—all in all, a perfect officer and gentleman (“Roland fut d’abord Officier, / Car il étoit bon Gentilhomme”).8 For the most part “Soldats François” won little praise; however, it may have provided inspiration for Rouget de Lisle when he composed his anthem “Roland à Roncevaux,” and it certainly drew renewed attention to the story.9 In 1814 Charles Nodier speculated in the Journal des débats about the possible survival of a fragment of the epic in some library, and in 1831 Chateaubriand suggested more specifically that fragments might survive in one of the former royal libraries.10 The search to recover the work had begun in earnest.
The earliest account of the story of Roland based on a specific medieval manuscript is that offered by Louis de Musset, marquis de Cogners, granduncle of the poet Alfred Musset, in his “Légende du bienheureux Roland, prince français” of 1817. Musset had access to what is now known as the Châteauroux manuscript and drew on it to retell the story of Charlemagne, Roland, and the Battle of Roncevaux.11 Guyot des Herbiers, a family relation, began to prepare an edition of this manuscript, but died in 1828 without having finished.12 It was not until 1832, however, when Louis-Henri Monin, a student at the Ecole Normale, published his Dissertation sur le roman de Roncevaux that a full version of the poem at last appeared in print. Monin based his edition on the Paris text (now Bibliothèque Nationale de France, MS Français 860), which he compared with the Châteauroux manuscript, using the transcription of Herbiers.
For some time, scholars had also been aware of the existence of an earlier version of a poem about Charlemagne and Roland in a manuscript in the Bodleian, although they were not quite sure what this poem was. It was known to Thomas Tyrwhitt, who appears to have read the entire work and mentions it in one of the notes to his Canterbury Tales of 1778; to Abbé Gervais de la Rue, who had worked in the Bodleian while in exile in England during the Terror; and to John Josias Conybeare, formerly professor of Anglo-Saxon at Oxford.13 De la Rue classified it as “un Roman de la bataille de Roncevaux qu’on appelle encore le Roman des douze pairs de France” and found it not without its interest, primarily because of its age. He even published a few excerpts in his Essais historiques sur les bardes, les jongleurs et les trouvères, and it is here that the famous opening lines first appear in print, although publication was delayed until 1834, years after de la Rue had examined the manuscript.14 But de la Rue never connected the Oxford poem to Taillefer; indeed, he lamented that he had never been able to find even a fragment of Taillefer’s song and poured scorn on those like Paulmy who claimed to have found traces of it in later romances.15 In short, when Michel came to Britain in 1835, both French and British scholars had been dreaming of discovering Taillefer’s lost performance for at least half a century, but nobody yet believed he had found it.
Guizot sent Michel to Britain as part of an extensive cultural mission to recuperate fragments of early French literature and history.16 As one of the many treatments of Charlemagne, the Oxford poem was numbered among these desired fragments and thus justified a trip up from London, where Michel was copying Gaimar’s Estoire des Engleis and Benoît de Sainte-Maure’s Chronique des ducs de Normandie.17 On 13 July Michel announced the discovery that would eclipse the rest of his voyage in a triumphant letter to one of his patrons, Louis-Jean-Nicolas Monmerqué, a member of the Académie des Inscriptions et Belles-Lettres, and Monmerqué quotes from this “cry of exultation that burst out at the moment of a discovery” in his own letter to Guizot a week later. Michel believed he had found not just a poem about Roland that was older than any of the others that had survived but something far more precious, a copy of the great lost chanson de geste of Taillefer itself:
I am writing to you from Alfred’s city, a few