Hans-Erich Keller

The Song of and its Audience

HIS PAPER WILL FOCUS upon the dating of the Oxford version of the Song Tof Roland, which I maintain is in all probability a relatively faithful copy of the hypothetical poem which Cesare Segre has labeled "omega" in his stemma. 1 I also wish to comment in particular upon the dating of the latter, the antecedent of the Digby 23 manuscript.

1. Notwithstanding objections on the part of many scholars, and without in any way precluding the existence of earlier versions consisting of a song about the battle of Roncevaux in the eleventh century which unfortunately do not survive today, I find it most difficult to avoid the conclusion that the text preserved at the Bodleian Library at Oxford dates from the last quarter of the twelfth century. To support this contention, I will purposely circum- vent the particular linguistic controversy which involved Jules Horrent and Ronald Walpole in the Fifties2 as well as the paleographical dispute which pitted Ian Short against Charles Samaran in the early Seventies,3 and will present new material which suggests a solution to the problem from three angles heretofore unexplored.

I would first like to offer a personal observation which I consider signifi- cant: while examining the oldest extant fragments of Wace's Roman de Brut, also preserved at Oxford, which already in 1898 were dated by the Keeper of Manuscripts there as being of the latter part of the twelfth century, I was struck by their very close paleographical similarity to Digby 23 and their affinity to a most important fragment of the Voyage de Saint Brendan which was probably written during the first half of the thirteenth century and is contained in the same collection at Oxford as the Brut fragments.4 To cite but just a few examples, all three manuscripts very frequently use an acute accent on the letter í (as in Roland, v. 69, dís 'ten', Brut, v. 7034, mís 'put', Brendan, v. 19, seínz 'saint'), or on ú with the meaning of 'or' (as, for example, in Roland v. 102, Brut v. 7074, Brendan v. 1510); in addition, it is always found on the preposition á; all three manu- scripts consistently employ the sign to indicate pre, per, par, or pur; and all three utilize the abbreviations or é for the vernacular form est, as found in Latin manuscripts. In 1974, a reexamination led the Assistant Keeper of Western Manuscripts of the same library to the conclusion that the Brut fragments were certainly written no later than the beginning of the thir- teenth century and could even be attributed to the late twelfth century.5 On

259 260 Olifant/Vol 6, Nos. 3 & 4/Spring & Summer 1979 the basis of the paleographical closeness to Digby 23 which can be re- marked in these fragments, it would appear that, at least from an im- pressionistic point of view, the Oxford Roland would seem not to be older than the last quarter of the twelfth century.

A second consideration derives from the orthography of the Oxford manuscript. Thanks to the research of Jules Horrent, we know that it is definitely tainted with Anglo-Normanisms.6 Several years ago I discovered that verse 3986, which reads "Truvee li unt le num de Juliane," had been erroneously corrected to truvét in most editions,7 despite the fact that Mildred K. Pope positively asserts that the form with -ee in the masculine of the past participle is one of the leading features of Anglo-Norman scripta, and one which can occasionally be found before the end of the twelfth century. Verse 3986 would thus suggest an additional indication that the Oxford Roland must be considerably younger than Charles Samaran is inclined to believe.

Finally, the German Ruolantes liet should be considered, although its date of composition, too, has been debated for more than a century. Most scholars would readily agree that this poem depends on a hypothetical version, Segre's intermediary type "beta," which is another offshoot of his "omega." Until recently, however, no one was positive as to when and where the priest Conrad composed his adaptation of into Middle High German. Evidence for a correct dating emerged very slowly, and I could only recently establish8 that the Ruolantes liet was in all probability composed at the court of Henry the Lion (Duke of Saxony and son-in-law of King Henry II Plantagenet) at Brunswick, not before 1180, and in most likelihood even after his return with his wife from England in 1185, following the exile which Barbarossa imposed upon him in 1180. The epilogue of the Ruolantes liet relates that the wife of Henry the Lion, Matilda (1156-1187), requested a translation of one of the versions of the French poem which was circulating in England at that time. The latter could only have been a "beta" version (see below); nevertheless, Conrad's indirect rendering of it into Middle High German via the medium of Latin is still the text which most closely resembles the Oxford Roland.

In the second half of the twelfth century, then, there were already at least two versions of the Song of Roland circulating in England: one of Segre's hypothetical type "alpha," from which Digby 23 derives, and one of a hypothetical "beta" version of which the earliest literary offshoot known today is the German Ruolantes liet. A primary difference between the two Keller/The Song of Roland and its Audience 261 hypothetical types, and perhaps even more of them which have been lost, lies in the fact that the "alpha" Roland must not only have contained the voluntary conversion of the Saracen queen , which none of the offshoots of the "beta" version contain, but also 's trial by the grand jury composed of 's barons. Of all of the "beta" ver- sions, only Conrad's adaptation mentions a conversion per se, but even he does not treat it as such. Another significant difference lies in the exclusive selection of the Christian name Juliana for Bramimonde in the Oxford Roland, a feature which must have derived from an "alpha" version since it does not figure in any of the derivatives of the hypothetical "beta" Roland: as I have noted elsewhere, the name Juliana is quite important inasmuch as Saint Juliana was particularly venerated in England.9

2. But what made the Song of Roland so popular in the kingdom of Henry II that even at that date it was known in England in at least two different versions, this despite the frequent mention of Saint Denis as well as the monastery of Saint-Denis? The Angevine circles were certainly delighted that it was Geoffrey of Anjou who carried the royal banner in the decisive battle against , the highest representative of the pagan world, and that the same Geoffrey even comforted the Emperor at the time of his greatest suffering (the scribe of the Oxford manuscript even introduces Henry II himself into the bargain in v. 2883 as Geoffrey's brother!]. Yet the interest of the Plantagenets in this poem lies elsewhere, as has been demonstrated long ago, for Horrent already observed in 1951 that "Le mystérieux Henri de la Chanson de Roland d'Oxford pourrait n'être que la projection dans le passé du régnant en Angleterre à l'époque de notre remanieur. Celui-ci aurait inventé de faire rejaillir sur un ancêtre imaginire de son roi de même nom que lui la gloire de Roncevaux."10 I could not be more in agreement, for this supports my contention that the Oxford Roland derives in all probability from Segre's "omega," a poem which achieved its popularity due to the political ambitions of Henry II, since they aimed precisely at the thaumaturgical function of kingship advocated by the poem—especially in its last part, the trial of Ganelon. In this respect, the hypothetical "alpha" and "beta" versions do not essen- tially differ; the popularity of both attests to their political actuality.

Let us then briefly recall the problems at stake near the end of the Oxford text when Charlemagne questions his barons as to whom loyalty is due while in the military service of the suzerain. It is significant that in the Song of Roland the sword had to decide, as in the old Germanic laws, but this could very well be an anachronistic feature since it was typical for the 262 Olifant/Vol. 6, Nos. 3 & 4/Spring & Summer 1979 atmosphere in which the "omega" poem was created. However, the court of Henry II was much clearer as to the question of loyalty: here every crime was spelled out. As the historian François-Louis Ganshof explains: La féodalité anglaise est une création de la Conquête de 1066; les relations féodo-vassaliques, telles qu'elles existaient dans le Duché de Normandie, ont été étendues par ses successeurs d'une manière plus complète que partout ailleurs, mais de façon é servir les desseins de la royauté. De manière plus complète qu'ailleurs, en ce que tout le sol fut approprié par la Couronne et que l'alleu, la pleine propriété, s'est trouvée complètement exclue; la "franche aumône" (angl. frankalmoin) elle-même, ailleurs alleu ecclésiastique privilégié, est en Angleterre considéré comme une tenure à charge de prières. Toute terre a constitué une tenure aboutissant directement ou indirectement au roi: il n'y a pas eu de , au sens français du mot (fief de chevalier, "vavassorie," "sergenterie") qui ne dépendît directement ou indirectement du roi.11 But, as Walther Kienast has remarked,12 as long as the Norman barons were at home on both sides of the Channel not every act of open disobedience against their king could be judged as high treason, for the monarch himself in his function as Duke of Normandy and ruler over vast territories in the west of France too often disobeyed his own overlord, the King of France; he therefore could hardly condemn his own in each instance. Kienast has established that the concept of treason in twelfth-century England was limited to specific crimes termed "treacheries": abandoning a castle to the enemy, liberating prisoners who had been incarcerated on the orders of the king, attacks against the king's life or the throne, attempts on the lives of his functionaries, and the like. Notwithstanding these limitations, it is evident that Henry II and his advisers must have been highly interested in how the trial of Ganelon was conducted and in its conclusion, the more so since the whole poem was of importance as an exemplum for the relationship be- tween Henry II and his vassals, dealing in particular with one who had concluded an alliance with the enemy. Was this high treason in the eyes of Henry II and his party—even though his adversaries might regard it as an act of non-treason in accordance with Norman-Angevine law in England— or did it constitute a feudal dispute in terms of French law? The result of the judicial duel between Thierry d'Argonne, the representative of royal cen- tralization, and , who may be seen as a symbol for the powerful barons, must have pleased the party of Henry II; this could explain the great success of the poems in England.

When it is recalled that Henry II instigated the murder of Thomas Becket Keller/The Song of Roland and its Audience 263 in 1170, a parallel between Ganelon and the Archbishop of Canterbury may be seen: each is representative of the old, feudalistic system; the indepen- dent actions of each vis à vis the throne could not be condoned. According to Charlemagne (by extension Louis VII, under the influence of Abbot Suger of Saint-Denis), every 's actions while in the service of the king were accountable to him, a position which progressively entrenched itself in the mind of Henry II with the passage of time to the point that he ultimately incarcerated his own wife, Eleanor of Aquitaine, unable to accept her rebellious actions and those of his sons, who would retain control of the western part of France in defiance of his centralistic policy.

At this point it is paramount to reiterate the fact that only the Oxford version of the Roland contains the council of Charlemagne's many barons (vv. 3780-3813), who assume the rôle of a grand jury in judgment of Ganelon and ultimately decide that nothing is to be gained from his punishment, to which the Emperor replies: "Vos estes mi felun" (v. 3814). We recall that it was Henry II who, weary of the necessity to serve in the function of ambulatory judge, instituted the system of jurors and jury in England which still today forms part of Germanic law on both sides of the Atlantic Ocean.

R. Howard Bloch attests that

Only with the renewed interest in Roman Law of the twelfth and thirteenth centuries did treason again become associated with the crime of laesae maiestatis, an infraction directed against the more abstract and extended political body of the state, or its represen- tative, the king. It was, moreover, not until the middle of the fourteenth century that war against the monarch constituted treason and that misdeeds of this type were officially considered political crimes against the state.13

3. Furthermore, it has been stressed long ago that the Anglo-Norman character of the Oxford manuscript is not strong enough to hide the fact that, purely from a scribal point of view, it must have been copied, directly or indirectly, from a continental model; indeed, numerous other factors point to a previous text composed in the surroundings of this Capetians. The question as to whom loyalty was due while in the military service of the suzerain was also central in the Capetian state of Kings Louis VI and VII and their powerful adviser, Abbot Suger of Saint-Denis. In his invaluable study on the Capetian government in the twelfth century, Eric Bournazel stresses that 264 Olifant/Vol. 6, Nos. 3 & 4/Spring & Summer 1979

Le royaume, après les difficiles décennies du Xle siècle, entre dans une ère de création, de tension, de bouleversements. Il connaît progressivement un essor économique, démographique, intellectuel dont les historiens s'accordent à constater l'ampleur tout en mesurant les limites. Au sein d'une société chrétienne, d'une société d'ordres, l'accord est loin d'être universel et les luttes politiques gardent une grande âpreté: les seigneurs, souvent durement frappés par l'évolution économique, n'entendent pas renoncer à leur puissance, s'efforcent de profiter encore du système féodo-vassalique et veulent maintenir leur independence. Certes la Royauté reste marquée du sceau du sacre qui lui confère un prestige incomparable; plus sûre d'elle et pourtant confrontée à de violentes oppositions, elle se doit d'adapter ses structures aux nécessités nouvelles, pour remplir ses engagements, pour être fidèle à sa mission.14

Nevertheless, as Joseph R. Strayer indicates,

[on the Continent] had far less of a hierarchical char- acter [than in England]; it revealed clearly that it had grown out of a series of improvisations required to meet immediate emergen- cies. In much of France and most of Germany it was possible for a lord to have vassals and yet to possess his lands as an allod, free from any feudal obligations to a superior. Thus there could be many small feudal groups which had no connection with each other and which escaped almost entirely from the control of higher authorities. Even where the feudal chain of command existed in theory, it might be broken in fact.15

Under these circumstances, Ganelon's trial assumes a particular impor- tance and suggests that this episode was conceived at a critical moment for the Capetian monarchy: such a period could very well have been the time of the Second Crusade, when many dignitaries of the kingdom who had not taken the Cross with their king were endeavoring to take advantage of his absence in order to revolt against the three regents: Raoul, Count of Vermandois (the king's first cousin once removed, who was excommuni- cated during most of this period for personal reasons); Archibishop Sam- son of Reims (whose heavy dependency on the Crown resulted from con- stant pressure exerted from the Count of Champagne, one of the mightiest barons of the kingdom who in addition had not accompanied Louis VII on the Crusade); and—most importantly—Abbot Suger of Saint-Denis, the great statesman and longtime adviser of the King and his father, Louis VI. The most crucial moment of that uprising was certainly the untimely return of Robert, Count of Dreux, the younger brother of King Louis VII who had left the Crusade with his men following one of his recurring disputes Keller/The Song of Roland and its Audience 265 with Louis, hoping to overthrow the king by his early return; it was only thanks to Suger's prestige and the influence of the French pope, Eugene III, that a major catastrophe could be averted. Ganelon's trial appears therefore to be the reflexion of an author who was well aware of what Suger tried to achieve in his long-standing political career, and of what would have been in jeopardy if it had not been for the Church. Aryeh Graboïs confirms that

la croisade de Louis VII eut des conséquences favorables sur le redressement du pouvoir royal, et cela malgré l'absence du roi pendant deux ans. En vertu du privilège de croisade, la régence fut confiée à l'abbé de Saint-Denis, Suger, qui était muni d'une sorte de vicariat pontifical. Puisque le royaume entier était mis pendant la croisade du roi sous la protection de l'Église, la paix de Dieu devint la paix du royaume; la réunion des deux notions fut comprise ainsi par les contemporains. La chronique de Morigny raconte comment, après avoir célébré le concile de Reims en 1148, Eugène III menaça d'anathème toute personne qui oserait violer la paix du royaume avant le retour du roi.16

Why then do the vassals of Charlemagne's council waver with respect to Ganelon's action? This is certainly an anachronism, as has recently been convincingly argued by Thomas N. Bisson, who affirms: "Political realities dictated alliances between the king and one or more of the major vassals, not the harmonious solidarity idealized (and perhaps lamented) by the Roland poet."17 But we must also consider vv. 3803-05 of the poem, in which the barons remind Charlemagne: "Morz est Rollant, ja mais nel revereiz;/ N'ert recuvret por or ne por aveir;/ Mult sereit fols ki ja se cumbatreit." Their behavior lends itself to a possible interpretation of the reasons behind Thierry's support of the king's cause when he says (vv. 3824-25): "Bels sire reis, ne vos dementez si!/ Ja savez vos que mult vos ai servit." This implies that Thierry d'Alsace, Count of Flanders, who in the poem is correctly portrayed as the brother (-in-law) of Geoffrey Planta- genet, had personal feudal ties with the Capetian monarchy, bonds which he himself stresses when he adds (v. 3826): "Par anceisurs dei jo tel plait tenir," a probable allusion to the fact that the allegiance of the counts of Flanders to the French Crown dated back to the time when the founder of the dynasty of Flanders eloped with Judith, daughter of Charles the Bald. Thierry, who had difficulties in establishing himself in Flanders, had thus strong motivation to elicit the king's favor to ensure his success in Flan- ders.

Ganelon's trial yields yet many other details, of which I would like to mention only two examples at this time, that corroborate my contention 266 Olifant/Vol. 6, Nos. 3 & 4/Spring & Summer 1979 that at least this part of Segre's hypothetical poem "omega" must have been composed about 1150. The first deals with the name of Thierry's opponent in the anachronistic judicial combat, "Pinabel del castel de Sorence" (v. 3783). On another occasion18 I have demonstrated that Sorence is a simple scribal error for Sorente, the present-day Sorrento near Naples. The name of this city could have remained fresh in the mind of the poet if he had taken part in the return of Louis VII and his wife Eleanor from their crusade in the early summer of 1149, when they met with King Roger II of Naples and the Sicilies in Potenza (Calabria) en route to Rome, since Sorrento fell to Roger only in 1133 after a long and fierce resistance—a fact which was certainly still remembered when Louis and Eleanor visited him. It was during this visit that the poet could have heard of a baron named Pinabel, although there is no proof that he was necessarily identical with the Pinabellus found by Emily Jamison in the Catalogus Baronum composed for King Roger about 1150 at Casaldianni, not far from Benevento.19 Indeed, the name Pinabel can be explained only by southern Italian linguistic elements, which I have identified elsewhere as meaning 'beautiful like a pine tree',20 although a French audience might not have recognized it as such and might easily have associated the name with the familiar human attribute, la pine.21 In any event, the creation of Pinabel, castellan of Sorrento, may be imputed to someone who had contact with the State of Roger II of Naples and the Sicilies, given the fact that Sorrento was incor- porated into that State only in 1133.

This does not necessarily suggest that the poet of the "omega" Roland was himself a member of the royal entourage at the time of Louis's meeting with King Roger: the poet might very well have heard of that visit and details respective to it after Louis's return to France late in 1149, perhaps from the king's chaplain, Odo of Deuil, who became Suger's successor as abbot of Saint-Denis in 1151. In his De profectione Ludovici VII in Orientem (written before his own death in 1162), Odo of Deuil describes the defeat of Louis's army by the Turks in a mountain pass of Anatolia, an event which for his contemporaries had to be strongly reminiscent of Charlemagne's catastrophe at Roncevaux. A poem which would indirectly explain Louis's disaster as being caused by treason, and thus somewhat attenuate it, could not be more welcome than immediately after the king's return to France just after 1149, when he had to account for the loss of so many men.

4. Further evidence for a composition of the "omega" Roland around 1150 is suggested by v. 3796 of the Oxford manuscript: "Icels d'Alverne i Keller/The Song of Roland and its Audience 267 sunt li plus curteis," a comment made during Ganelon's trial, after which the poet stresses that "Pur Pinabel se cuntienent plus quei" (v. 3797). This would seem to be an allusion to the intervention of Louis VI in the affairs of Auvergne in 1126 in favor of the bishop of Clermont-Ferrand, who had accused the Count of Auvergne of having violated the peace of his bis- hopric (the treuga Dei) and had forced the king to summon the count to his court. This count was a personage no less important than the father of Eleanor, William X, duke of Aquitaine, who was then judged at Louis's court in the presence of the king's immediate vassals. Verses 3796 and 3797 may then be seen as a warning directed at Queen Eleanor as an exemplum par excellence for others that an independent feudal policy on her part with regard to her territories would no longer be tolerated by her husband, the king.

Pinabel, Ganelon, and their relatives thus appear as symbols of the feudal force which tried to bind the local dynasty to the domain over which it ruled: in other words, they are representative of the old Germanic system. It is therefore only logical that the barons of Auvergne, who had been forced to submit to the centralization of the North, would sit back passively and "hold their breath." Such a supposition also explains why the poet who reworked an earlier version of the events of Roncevaux around 1150 and added the second part calls the Auvergnats the most curteis. Glyn S. Burgess22 has shown that the military virtues of curteis began to be coupled with those of politeness, sociability, wisdom, and even physical beauty under the influence of the South and Eleanor's entourage in particular; this fact again indicates the years around 1150 in northern France, for Burgess finds these notions first in the Roman de Thèbes.23 Obviously, the Auvergnats had many reasons for demonstrating restraint—and not only because they were the southernmost subjects of the kings of France and, as a result of their geographic location, were strongly influenced by the civilization of the Midi. In addition, the First Crusade had been decided at Le Puy; Louis VI then had to intervene twice in their internal affairs (in 1122 and 1126); finally, the Second Crusade was forced upon Suger at Bourges, the capital of the same ecclesiastical province to which Auvergne belonged.

5. Evidence would then seem to point repeatedly to the years immediate- ly following the return of Louis VII from the Second Crusade, when he found his kingdom more centralized than ever and, in the words of Grabois, "une autorité royale accrue qui, pendant la régence, était munie des pouvoirs laïques comme des ecclésiastiques," thanks to its mastermind, Abbot Suger of Saint Denis, the most influential current regent. Conse- 268 Olifant/Vol. 6, Nos. 3 & 4/Spring & Summer 1979 quently, Suger's rôle in the composition of the "omega" Roland simply cannot be stressed enough.24 I would like to mention only three facts at this time which strongly suggest that the first part of Segre's hypothetical "omega" version was reworked, and the second part composed, by an author directly inspired by Suger just before his death in January 1151.

First the interest of the monastery of Saint-Denis in the Rolandian matter, which has also recently been emphasized by Gabrielle M. Spiegel,25 must be mentioned. In enumerating the relics contained in 's pommel, only the name of Saint Denis is preceded by the qualification "mun sei- gnur" (v. 2347), a title accorded by the people to its favorite saint since the time of Wace: thus since the same period.26 It is also significant that Margariz of Seville boasts after Ganelon's treason (v. 973) that the Saracens will occupy the whole territory of France within one year, going as far as to say "gesir porrum el bure de saint Denise," an indication that the first part of the poem was reworked, and that the poet of the "omega" Roland already considered the monastery of Saint-Denis as the center of France at the time of the remaniement.

Secondly, it must be recalled that Saint-Denis, under Suger's instigation, began to compete with during this same period and in particular following the Second Crusade in order to declare Charlemagne a saint. Although Aachen ultimately triumphed and obtained the sanctification of Charlemagne in 1165, Saint-Denis remained faithful to the cult of the great Emperor and always cherished his memory, as is demonstrated by the propaganda which was made by the same monastery for Philip Augustus as another Charlemagne half a century later. Certainly the name of the royal banner, oriflamme,27 would seem to stand as a striking testimonial to this effect, for, in spite of the fact that Pope Leo III presented Charlemagne with a little banner in the year 800, it has been historically proven that Suger was instrumental in establishing the scarlet, notched banner of the counts of Vexin, whose territory had been administered since 1077 by the abbots of Saint-Denis (following the extinction of this family), as a royal symbol. The banner was presented to the Capetian king at Saint-Denis by Pope Eugene III himself on June 11, 1147, in the presence of the entire royal family including Queen Eleanor, as the new official banner of the French king- dom. This explains why the poet of the "omega" version was able to state that the banner had first been Saint Peter's and was therefore called Romaine, but that from the French battle cry it derived its new name of Munjoie (vv. 3093-95), further evidence for a date of composition of the "omega" Roland during or after the period of the Second Crusade. Keller/The Song of Roland and its Audience 269

Finally, the fact that Suger was a brilliant historiographer cannot be ignored: he himself wrote the Vita Ludovici Grossi Regis (ca. 1144), and very likely had asked Odo of Deuil to take notes on the Second Crusade for him while he remained behind as a regent of France. His bibliographer William relates that, a tireless storyteller, Suger kept his monks awake and, "narrabat vero, ut erat iocundissimus, nunc sua, nunc aliorum, quae vel vidisset vel didicisset gesta virorum fortium, aliquotiens usque ad noctis medium"28 (italics mine). He also knew the history of the kings of France by heart, and was responsible for the reestablishment of Saint-Denis as a depository of the archives of the monarchy, thereby becoming the founder of the Grandes Chroniques de France.

6. Given the arguments presented in this study, which can be substan- tiated by many others, I contend that the "omega" version, from which the English model of the Oxford manuscript (Segre's "alpha") derived, was composed around 1150 at the Abbey of Saint-Denis for the cause of the Capetians, under the inspiration of Suger.

7. What conclusions can be drawn from these observations with respect to the genesis of the Song of Roland? Given Suger's predisposition toward tradition, which for lack of space we could here see exemplified only in the case of the judicial duel and the oriflamme, I maintain that his story of the tragedy is based on an older eleventh-century poem which very likely contained the battle of Roncevaux and the comradeship of Roland, , and Turpin, as well as Ganelon's betrayal. This is also suggested by icono- graphical data unearthed by Jacques Stiennon and Rita Lejeune, who relate in their opus magnum that the first representation of the narrative from Ganelon's treason to Roland's death is found on the tympanum of the cathedral of Angoulême in western France, the consecration of which can be dated to about 1123. It has further been established that as early as 1131 the name of the traitor Ganelon is already mentioned in the cathedral of Nepi, not far from Sutri, a city which was about one day's journey north of Rome for a pilgrim.29 However—and this is crucial to my position—the first iconographical indications of the second part of the poem are found only in the illustrations of the Heidelberg manuscript of the Ruolantes liet, which has been dated toward the end of the twelfth century.

Furthermore, it is extremely significant that the name Baligant has been identified by a French Islamist as Yahya ben Ghâniya,30 a leader of the Spanish Moors who in 1134 inflicted a crushing defeat at Fraga on the Christians of Aragon and their French allies under Alfonso I the "Batalla- 270 Olifant/Vol. 6, Nos. 3 & 4/Spring & Summer 1979 dor," who was mortally wounded in this clash, an event which therefore forms a definite terminus a quo for all the versions of the Song of Roland known to us today including those of the Pseudo-Turpin. In previous research I have even ventured a comparison of the "omega" Roland to Geoffrey of Monmouth's Historia Regum Britanniae, where Arthur takes a late revenge on the Romans at Siesia in Burgundy for their victory at Alesia over the hero of the last Celtic resistance in Gaul, Vercingetorix.31 Unfor- tunately, lack of space again prevents a recapitulation at this time of the very strong influence of Geoffrey of Monmouth's model for this work, definitely composed prior to 1136,32 on the "omega" version of the Song of Roland.33

The text of the second part of the Digby 23 manuscript—in other words, everything following the death of the twelve peers at Roncevaux and Charlemagne's pursuit of the Saracens to the Ebro (with the exception of certain interpolations in the first part which were necessary because of the reworking)—is then an echo of a mid-twelfth century poem: a poem which enhances the tragedy that befell Roland through the use of a drama center- ing around Charlemagne, in order to further Capetian interests as formula- ted by the political genius of Abbot Suger. In conclusion, it seems appro- priate to reiterate what I have said on an earlier occasion: the Oxford version of the Song of Roland reflects the quintessence of the national spirit of the Capetian kingdom toward the middle of the twelfth century, of which Suger's Saint-Denis was the soul.34

Hans-Erich Keller The Ohio State University

1Cesare Segre, ed., La Chanson de Roland; Documenti di filologia, 16 (Milan and Naples: Ricciardi, 1971), p. xiv.

2Jules Horrent, La Chanson de Roland dans les littératures française et espagnole au moyen âge; Bibliothèque de la Faculté' de Philosophie et Lettres de l'Université' de Liège, 120 (Paris: Les Belles Lettres, 1951), pp. 34-42, and Ronald Walpole's review of Horrent's book in Romance Philology, 6 (1952-53), 337-48. 3Ian Short, "The Oxford Manuscript of the Chanson de Roland: A Paléographic Note," Romania, 94 (1973), 221-31; Charles Samaran, "Sur la date approximative du Roland d'Oxford," Romania, 94 (1973), 523-27. Keller/The Song of Roland and its Audience 271

4Hans-Erich Keller, "Les Fragments oxoniens du Roman de Brut de Wace," in Mélanges offerts à Carl Theodor Gossen, Vol 1 (Berne and Liège: Francke-Marche Romane, 1976), pp. 453-67. Prof. Rupert T. Pickens has informed me that the same paleographical features also occur in the oldest manuscripts of Philippe de Thaün's work, dated from the beginning of the thirteenth century. 5Letter of September 16,1974, from Mr. Bruce C. Barker-Benfield to the author of this study. 6Op. cit., pp. 34-42. 7This is not the case, however, in the most recent edition, that of Gerard J. Brault, The Song of Roland: An Analytical Edition. Vol II: Oxford Text and English Translation (University Park-London: The Pennsylvania State University Press, 1978), p. 242. See Mildred K. Pope, From Latin to Modern French with especial Consideration of Anglo-Norman (Manchester: The University Press, 1934; rev. ed. 1952), par. 1235: "According to Vising the use of this device is not frequent before the fourteenth century, although sporadic instances of -ee for -e occur from the later twelfth century on (Purg. Patr., p. 11)." 8Hans-Erich Keller, "Der Pfaffe Konrad am Hofe von Braunschweig," in Wege der Worte. Festschrift für Wolfgang Fleischhauer, ed. D. C. Riechel (Cologne: Böhlau, 1978), pp. 143-66. 9See Hans-Erich Keller, "La Conversion de Bramimonde," Olifant, 1 (October 1973) 1, pp. 3-22; reprinted in Société Rencesvals pour l'étude des épopées romanes. Vie Congrès International (Aix-en-Provence, 29 Août - 4 Septembre 1973), Actes, pp. 190-91. Gerard J. Brault, op. cit., Vol. 1: Introduction and Commentary, p. 476, states that I base my argumentation entirely on the provenance of several manuscripts containing saints' lives including that of Juliana, and that I conclude from this that England was a more important center of this saint's cult than the Spanish monastery of Santillana in Asturia. This does not take into account my consideration of the monumental study by Karl-Ernst Geith, Priester Arnolts Legende von der Heiligen Juliana. Untersuchungen zur lateinischen Juliana- Legende und zum Text des deutschen Gedichtes (diss. Freiburg i. Br., 1965), p. 17, in which Geith establishes beyond any doubt that the cult of Saint Juliana was trans- ferred from Naples to Canterbury in the second half of the seventh century (see Karl-Ernst Geith, op.cit., pp. 14, 99ff.; see also Wilhelm Levison, England and the Continent in the Eighth Century [Oxford: Clarendon Press, 1946], p. 141). The oldest vernacular poem concerning Saint Juliana is that of Cynewulf in Anglo-Saxon of the beginning of the ninth century, which was followed by more than half a dozen vernacular versions in English throughout the ; the French Vie de Sainte Julienne (dated after 1150) was probably also composed in England (see my study "La Conversion de Bramimonde," note 64). Brault maintains that I only add a thirteenth-century illustration produced at Saint-Albans to the iconographical corpus concerning this saint, but in my article (Actes, p. 192) I stress that I selected this illustration at random ("entre autres") and that many others are found in medieval texts produced in England. 272 Olifant/Vol. 6, Nos. 3 & 4/Spring & Summer 1979

In his paper "La Légende de sainte Julienne et ses rapports avec la Chanson de Roland" presented at the Seventh International Congress of the Société Rencesvals held at Liège in 1976 (see Les Congrès et Colloques de l'Université de Liège, vol. 76: Charlemagne et l'épopée romane, vol. 2, pp. 547-63), Paul E. Barrette pays great attention to Geith's findings and follows up his leads very carefully, arriving at a thorough history and description of the Saint Juliana legend in medieval European literature. The main thrust of his study is to prove that England does not occupy any particular place in the cult; he insists rather on the interest in this saint at her burial place in Campagna, north of Naples; Barrette even ventures to ask himself if the same person who recorded the name of Pinabel on the occasion of Louis VII's visit to King Roger of Naples and the Sicilies (see below) could have also learned of Saint Juliana in southern Italy (p. 558), which implies that he thinks it possible that the episode of Bramimonde's conversion already belongs to the Capetian version of about 1150 (see below). This view, however, overlooks the circumstances which made the cult of Saint Juliana particularly popular in England. In my article "La Conversion de Bramimonde" (Actes, p. 190), I point to the fact that, between the fifth and the eighth centuries, two centers of this cult are found: one in southern Italy (north of Naples), the other in England, because the cult was transplanted to England in the seventh century when Hadrianus became abbot of Saint Peter and Paul at Canterbury and director of the cathedral school in 670. I further stress that, according to the Venerable Bede (Hist. Eccles., IV, I), Hadrianus had been an abbot in the very monastery near Naples in which Saint Juliana was venerated before he was transferred to England and took with him copies of saints' lives. Saint Juliana very likely figured prominently among these manuscripts: Bede himself included her life in his martyrologium. Furthermore, Cynewulf s poem, as the oldest in vernacu- lar about the saint, was based on one of the numerous Latin manuscripts written in England. England was thus the center of this cult north of the Alps from the seventh century and thereafter, without interruption. Barrette also does not take into con- sideration that the choice of the name Juliana occurs exclusively in the Oxford Roland, thus in a manuscript written in England, nor does he consider the fact that Bramimonde's conversion per se by its voluntary character alone points to the second half of the twelfth century, when such a thing became conceivable. There is simply no evidence whatsoever that this laisse was already contained in an earlier version. The scribe of the Oxford manuscript (or his immediate Anglo-Norman model, Segre's "alpha") must have added it to the text in remembrance of the verses 3679-81 (ed. Segre): "M[unt]et li reis e si hume trestuz/ E Brami[mund]e meinet en sa prisun:/ Mais n'ad talent li facet se bien nun." 10Jules Horrent, op. cit.. p. 324. 11François-Louis Ganshof, Qu'est-ce que la féodalité? (Brussels-Neuchâtel: J. Lebègue-La Baconnière, 1947), p. 188.

12Untertaneneid und Treuvorbehalt in Frankreich und England; Studien zur vergleichenden Verfassungsgeschichte des Mittelalters (Weimar: Böhlau, 1972), p. 295.

13R. Howard Bloch. Medieval French Literature and Law (Berkely-Los Angeles- London: University of California Press, 1977), p. 40. Keller/The Song of Roland and its Audience 273

14Le Gouvernement capétien au XIIe siècle, 1108-1180. Structures sociales et mutations institutionnelles. Publications de la Faculté de Droit et des Sciences Économiques de l'Université de Limoges (Paris: P.U.F., 1975), p. 1.

15Medieval Statecraft and the Perspectives of History (Princeton: Princeton Uni- versity Press, 1971), p. 83.

16In Mélanges offerts à René Crozet à l'occasion de son soixante-dixième anniver- saire, ed. Pierre Gallais and Yves-Jean Riou, vol. 1 (Poitiers: Société d'Études Médiévales, 1966), p. 592.

17"The Problem of Feudal Monarchy: Aragon, Catalonia, and France," Speculum, 53 (1978-1979), 478.

l8"The Song of Roland: A Mid-Twelfth Century Song of Propaganda for the Capetian Kingdom," Olifant, 3 (1975-76), 252-53.

19"Notes on S. Maria della Strada at Matrice, its History and Sculpture," Papers of the British School at Rome, 14 (1938), 71.

20Olifant, 3 (1975-76), 252 and n. 37 (with bibliography).

21See Walther von Wartburg, Französisches Etymologisches Wörterbuch, vol. 8, p. 550, col. 1.

22Contribution à l'étude du vocabulaire pré-courtois, Publications romanes et françaises, vol. 110 (Geneva: Droz, 1970), pp. 20-34.

23Op. cit., p. 23.

24This was the subject of a previous study which is crucial to the understanding of my position, "La Version dionysienne de la Chanson de Roland," in Philologica Romanica. Erhard Lommatzsch gewidmet, ed. Manfred Bambeck and Hans Helmut Christmann (Munich: Wilhelm Fink, 1975), pp. 257-87.

25See her important article "The Cult of Saint Denis and Capetian Kingship," Journal of Medieval History, 1 (1975), 43-69. 26See Walther von Wartburg, op. cit., vol.11, p. 449, col. 1. 27See Hans-Erich Keller, "La Version dionysienne de la Chanson de Roland," pp. 270-75, where the problem of oriflamme is treated extensively. 28Guillaume, Sugerii Vita, ed. A. Lecoy de la Marche, in Oeuvres complètes de Suger (Paris. 1967), p. 394. 29See Pio Rajna, "Un'iscrizione nepesina del 1131," Archivio Storico Italiano, ser. 4. vol. 18 (1886), 332. 274 Olifant/Vol 6, Nos. 3 & 4/Spring & Summer 1979

30Jean Poncet, "La Chanson de Roland à la lumière de l'histoire: vérité de Baligant," Actes du Deuxième Congrès International d'Études Nord-Africaines, Aix-en-Provence 1968 [Revue de l'Occident et de la Méditerranée, 1970, numéro spécial], especially pp. 131-32.

31See Hans-Erich Keller, "Two Toponymical Problems in Geoffrey of Monmouth and Wace: Estrusia and Siesia," Speculum, 49, (1974), 693.

32See Hans-Erich Keller, "Wace et Geoffrey de Monmouth: Problème de la chrono- logie des sources," Romania, 98 (1977), 1-14.

33See Hans-Erich Keller, "La Version dionysienne de la Chanson de Roland," pp. 278-80.

34"A Mid-Twelfth Century Song of Propaganda for the Capetian Kingdom," p. 285: see also Leonard Olschki, Der ideale Mittelpunkt Frankreichs im Mittelalter in Wirklichkeit und Dichtung (Heidelberg: C. Winter, 1913).