SOURCE: Merwin, W. S. Introduction to The Song of Roland, translated by W. S. Merwin, pp. v-xiii. New York: Vintage Books, 1970.

[In the following excerpt, Merwin outlines Charlemagne's disastrous battle in the Pyrenees in which Roland was killed and describes how the tale of the defeat, through retelling, eventually achieved legendary status.]

Some time near the end of July, Charles (Charles the King, Charles the Emperor, Charles the Great, Charlemagne) turned his army north toward the Pyrenees and France. The year was 778. He was thirty-six years old and he was not used to failure, but even the royal chroniclers would have difficulty in trying to describe his ambitious summer campaign in Spain as though it had been a success.

It had not been hastily conceived. Suleiman, the Moorish governor of Barcelona, had visited Charles in the spring of 777 to urge him to cross the Pyrenees, and the request, and Charles' response to it, were both influenced by dynastic and religious promptings which had histories of their own.

Suleiman was a member of the Abassid dynasty, descended from an uncle of Mohammed. Earlier in the century the Abassids had overthrown the reigning Umayyad dynasty and assassinated every member of it except one, Abdur Rahman, who had escaped to Spain and established himself there as the Emir. Suleiman's hatred of Rahman was understandable, and it had already led him to seek and to obtain the protection of his Christian neighbor, King Pepin of France, Charles' father.

There were other reasons why Charles would have been sympathetic to Suleiman. He was himself a member of a young dynasty, a matter of subtle importance in a world governed to a great degree by tradition. And then, Abdur Rahman, as the last representative of the Umayyads, stood for the family which, half a century before, had commanded the great Moorish invasion of France. At that time the apparently invincible Umayyads had forced their way as far north as Tours before Charles' grandfather, Charles Martel, turned them back. It was the Umayyads whom Charles' father, Pepin, had fought and at last driven from France.

But doubtless none of these considerations would have impelled Charles to cross the Pyrenees if it had not been for a more powerful and obvious motive: his own ambition. In the first nine years of his reign he had conquered Aquitaine, beaten the Saxons and the Lombards, and become the official guardian of Christendom, whose boundaries he had extended to the north and east. An expedition into Spain would give him a chance to unify the different parts of his realm in a common effort, and incidentally to conquer the as yet unsubjected Basque provinces. Suleiman probably stressed the apparent fact that Rahman was a menace to Charles' southern frontier, and very possibly he would have told the French king that if he were to attack Rahman now he could not help succeeding, that the Abassids themselves were raising an army of Berbers to send against the Umayyad, and that the people of Spain were on the point of rebellion. The exact details of the embassage and the terms of the agreement that was reached are not known. But by Easter 778 Charles was in Poitou with an immense army recruited from every part of his kingdom: it included Goths, contingents from Septimania and Provence, Austrasians, Neustrians, Lombards, Burgundians, and Bavarians. After Easter he crossed the western end of the Pyrenees, through the Basque country, at the head of half his army. He sent the other half around the eastern end of the mountains. They were to meet before Saragossa.

Just what happened that summer was carefully obscured in the accounts and will never be known. Certainly there were no great triumphs. The Christian natives of Spain did not hasten to overthrow the tolerant Moorish rule and welcome the Franks; on the contrary, the Christians of the kingdom of Asturias preferred their own independence to the presence of a foreign army however dear to the Pope. It is also possible that they were in league with Rahman. At any rate they resisted the Franks. The Christian city of Pampelona refused entry to Charles and had to be stormed; it was the only city in the entire campaign which was actually taken. The native rebellion against Rahman never amounted to much and Suleiman himself had a falling out with his Moorish allies on the African continent. When the Frankish army assembled before Saragossa the city defied it, despite Suleiman's diplomatic efforts; it is not known how hard Charles tried to take it, but he had no siege machinery, and he failed. By some time in July he had received the formal surrender of a few cities—a gesture which may have owed as much to his alliance with Suleiman as it did to his own army—and he had gained some hostages, and little else. There is no way of knowing just why he abandoned the campaign so early in the summer. It is possible that he saw nothing to be gained by staying, in the circumstances, and was simply cutting his losses. Supplies may have run dangerously low. It is conceivable that the campaign had turned out far worse than the accounts would lead us to suppose, and that the army was in fact retreating. Even if that were so it cannot have been a rushed or disorderly retreat: in August the army stopped at Pampelona long enough to raze the walls of the city to punish the inhabitants for their resistance, and no doubt to weaken the Spanish side of the frontier. It has been suggested (by Fawtier) that if Charles had not been in a hurry, for some reason, he would have paused long enough to celebrate the important feast of the Dormition of the Virgin on August 15th. At any event he did not do so, but pushed on into the Pyrenees.

What happened next is one of the great riddles.

In the earliest history of Charles' expedition, the one included in a chronicle known as the Annales Royales, there is no reference to any military action whatever in the Pyrenees. All later writers on the subject have agreed that the author had something of importance to be silent about. Of such importance, in fact, that his immediate successors evidently felt that mere silence would not serve to conceal it, and set about explaining it. The original Annales were rewritten and expanded roughly a quarter of a century after they were first compiled. It was long thought that the rewriting was done by Charlemagne's biographer Einhard, and though it is now certain that the changes are not his, the second edition of the chronicle is still referred to as the Annales dites d'Einhard. In this work there is a brief and contradictory account of something which happened on the way back from Spain. The Basques, it says here, from positions at the tops of the mountains, attacked the rear guard and put the whole army in disorder; the Franks were caught at a disadvantage and did badly; most of the commanders of the different sections of the army were killed, and the enemy, helped by the nature of the terrain, managed to carry off the baggage and escape. There is a reference, too, to the bitterness of Charles' grief.

Then there is Einhard's own account. In the first place he is more ingenious than his predecessors at making it sound as though the Spanish campaign had been a success; then, having built up the picture, he sets against it the Pyrenean ambush on the way back as a relatively minor mishap. It was the treacherous Gascons, he says; they waited until the army was spread out in a long line in the gorges, and then they rushed down and threw the baggage train and the rear guard into confusion. There was a battle in the valley and the Franks were thrown back. The Gascons killed their opponents, the rear guard, to a man, seized the baggage, and scattered under cover of night. Their flight was made easier by their light armor and the nature of the terrain. And then Einhard says, “In this battle Egginhard the royal seneschal, Anselm the Count of the Palace, and Hruodland, the Warden of the Breton Marches, were killed, with very many others.” It is one of the only two glimpses in history of the knight whose name would come to evoke one of the richest bodies of legend in the Middle Ages, and one of its greatest poems. The other is a coin, worn, but still displaying on one side the name Carlus, and on the reverse, Rodlan.

One final mention of the battle, by the chroniclers, is of interest. While the army was making its way back from Spain, Charlemagne's wife, in France, gave birth to a son, Louis, who would be his heir. Sixty years after the battle Louis' own biographer, a writer known as The Astronome, in speaking of it said that the names of those who fell in that action were so well known that there was no need to repeat them.

Of all the battles of the period, this one probably has excited most curiosity, and almost nothing about it is definitely known. It is not mere historical interest in the sources of the Roland story which still draws the speculation of scholars to what scanty evidence has come down to our times. In this case the theories of how the legend developed from the event are even more than usually dependent upon a notion of what the event was: a bitter but militarily unimportant misfortune, on the one hand, or one of the critical defeats of Charlemagne's reign, on the other.

Bedier, one of the great students of medieval literature in modern times and the editor of the Oxford text of La Chanson de Roland, propounded the theory of the development of the legend which was generally accepted for years. The battle, he believed, was a minor event which had been remembered in the area near the battlefield and had become a local legend; from those beginnings it had been retold and developed in monasteries and pilgrim sanctuaries along the route leading to Santiago de Compostella, in Spain; the route crossed the Pyrenees at Roncevaux—the Roncesvalles associated with the Roland story. Bedier, incidentally, was convinced that a number of the French chansons de geste developed in more or less the same way and may have been written by monks, or at least in collaboration with monks. With reference to the Roland, in particular, he cites the fact that the pass at Roncevaux was commended for admiration (complete with a monumental cross said to be Carolingian and other relics claiming descent from Roland and the battle) by the monks at Roncevaux in the twelfth century; he points out that one variant of the Roland legend is contained in a twelfth-century guide written for the benefit of pilgrims to Santiago de Compostella.

Bedier's theory was published just before World War I. It was subjected to criticism in the following decades by a number of scholars; one of the most interesting countertheories was put forward by Fawtier (La Chanson de Roland) in 1933. Fawtier analyzes the chroniclers' references to the battle and bases his conclusions, in great part, on the weaknesses in their accounts. The chroniclers, he insists, cannot have it both ways. Was it merely a massacre of the rear guard, or did it in fact involve the whole army and “throw it into disorder”? He poses some other interesting questions. Why, for instance, should the baggage train have been at the rear of the march, when it was usual to have it in the middle, especially in mountain country? Why should so many of the leaders of the different sections of the army have been in the rear guard (of course the legend itself, with its story of the Ganelon-Roland dispute, answers this one, but the legend in its final form came much later and a great part of it is concerned with the peculiar drama of this very situation). How many of these details, and how much of the picture of the lightning raid from the mountain tops may have been attempts to minimize and explain away a terrible defeat which had happened while Charles himself was in command?

In Fawtier's view, the battle, whether it took place at Roncevaux or elsewhere, was one of the great disasters of Charlemagne's career. The army, hurrying into the Pyrenees, was caught in a classical ambush: the van was blocked, the rear was then attacked, and the Franks had to fight their way forward, section by section, suffering losses so appalling that Charles never really managed to reassemble the survivors on the other side of the mountains, and instead set about hastily reorganizing the strong points in Aquitaine as though he expected further troubles from Spain. In fact the magnitude of the defeat was one of the things about the action which caught the popular imagination and contributed to the growth of the legend around the heroic figure of the doomed commander of the rear guard, Hruodland, Rodlan, Roland.

The legend may have grown in the region around Roncevaux, but it was elaborated in other parts of the kingdom too. By the late eleventh century, when the poem was written, it was possible for the poet to display, without fear of correction, an ignorance of the geography of Spain and, for that matter, of southern France, which indicates not only that he himself came from somewhere far from that part of the world, but also that the story and its heroes had long been familiar in places remote from the original battlefield. An audience at Roncevaux might just have been able to go along with the poet's assumption that Córdoba was near the hill city of Saragossa, which in turn was on the sea; it is unlikely that, even in the Middle Ages when simple experience was so meek an authority, they would have heard without a murmur that Narbonne and Bordeaux both lay on the same road north from Roncevaux. Furthermore, this shows a total ignorance of the Santiago pilgrim route and its monasteries, an interesting fact in view of the theory that the poem was composed in one of those places, on that route.

In Fawtier's opinion the story of the defeat was carried across France by its veterans, and in various localities, as it took on the character of legend through repetition, it was cast, in whole or in part, into the form of ballads. It is true that none of these survive, but then very little of the popular literature of the time has survived. The monks had nothing to do with the composition of La Chanson de Roland itself (although two other, later variants of the legend were composed by clerics). On the contrary, it was the legend, and perhaps the poem itself, which prompted the ecclesiastics at Roncevaux to exploit the pass as a pilgrim attraction—an enterprise which may have contributed to the poem's preservation.

There has been considerable controversy as to just when La Chanson de Roland was written. It must have been some time in the latter half of the eleventh century, but it is not possible to be much more definite than that. The poem apparently was already well known in 1096 when, at the Council of Clermont, Pope Urban II made use of it in his appeal to the chivalry of France to follow in the steps of Charlemagne and send an army against Islam. Many of the crusaders who responded to Urban's summons, and many who came later, must have been following an image of themselves which derived, at least in part, from the legendary last battle of the now transfigured Hruodland.

The poem, in its original form, has not survived. Modern knowledge of it is confined to six different versions, whose separate relations to the original are not plain. There is, for instance, a twelfth-century German translation by a Bavarian priest named Konrad. There is a Norse translation of the thirteenth century. There is a version in Franco-Italian, in the library of San Marco in Venice, which ends differently from the others. And there are three versions in French. One of them, known as Recension O, or the Oxford version, has survived in a single copy, Digby Mss 23, at the Bodleian Library, Oxford. It is supposed that it was a jongleur's copy of the poem. It is the oldest of all the versions, the most beautiful, and must have been much the closest to the original. Bedier's famous edition of the poem (the one I have used in making my translation) is based on the Oxford version, which Bedier compares at all points with the others.

Two other medieval retellings of the Roland legend are extant. One of them, the so-called Pseudo-Turpin, comes from Book IV of the twelfth-century Guide to the Pilgrims of Santiago de Compostella, to which I have already referred. It is in Latin prose and purports to have been written by the Archbishop Turpin himself. This worthy, as here presented, was with Charlemagne when Roland was attacked, and he had a vision in which he saw the soul of King Marsiliun being carried off by demons and the soul of Roland by angels. The narrative is clumsy, ill written, and encumbered with theological baggage. The other variant of the story, the Carmen de prodicione Guenonis, is also in Latin prose, but is shorter and more vigorous; it is possible that it is a translation of a lost French poem. A great deal of attention is paid to the character and actions of Ganelon. These two accounts, and the six surviving descendants of the Chanson itself were compared by Gaston Paris, who concluded that the author of the Pseudo-Turpin knew the Chanson but that the author of the Chanson did not know the Pseudo-Turpin variant; that there was no evidence of any relationship between the Pseudo-Turpin and the Carmen; that there was no way of establishing any relationship between the Carmen and the Chanson.

No decision about the spelling of characters and places could have satisfied everyone, and between the two extremes of modernizing and Anglicizing everything, on the one hand, or of keeping to the medieval versions in every case, on the other, I have not even been consistent. It would have struck me as affected and pointlessly archaic to have insisted on the original versions of names which have become familiar in modern English—Roland, Charles, Ganelon, Reims, Bordeaux. The work, after all, is a translation to begin with. But with names which, in my judgment, had not acquired such familiarity, I have either followed one of the original versions (sometimes there are several: Marsile, Marsilies, Marsilie, Marsiliun, Naimun, Naimon, Neimes, Naimes) or Bedier's standardized modern French version (Blancandrin, Balaquer, Thierry, Seurin), depending on which seemed preferable in the circumstances.

The Chanson de Roland, as it has survived in the Oxford version, consists of just under 4000 lines, arranged in laisses, or groups of lines all ending on the same assonance. The metrical pattern is based on a ten-syllable line with a clear strong beat. There are several drawbacks to trying to reproduce anything of the kind in English. For one thing, the assonance patterns: English is far more meager than are the Romance languages in the number of similar assonances which can be found for any given word ending. There have been translations of La Chanson de Roland which have aimed at producing assonance patterns like those in the original, but the results have been gnarled, impacted, and stunted, as the original certainly is not. It would also have been possible—and this too has been done—to translate the poem into a ten-syllable line more or less resembling that of the original. The trouble is that the associations of the ten-syllable line in English are not at all what they are in French. It would have been very difficult not to invoke the tradition of iambic pentameter in English literature, a gallery of connotations which would not only have been irrelevant to the poem but which also could not help disguising it. This is quite apart from my own strong disposition against even reading another transposition of La Chanson de Roland, or most anything else, into a sort of blankish verse.

I am not questioning the splendor of the verse in the Oxford version, the magnificence of the noise it makes. It would be boorish of me to do so after the pleasure they have given me. But the qualities of the poem which finally claim me are all related to a certain limpidity not only in the language and the story but in the imagination behind them, to a clarity at once simple and formal, excited and cool, to characteristics which I find myself trying to describe in terms of light and water. These qualities obviously could not be reproduced in any translation but I wanted to suggest them, and it seemed to me that I should try to do it in prose.