SOURCE: Goldin, Frederick. Introduction to The Song of Roland, translated by Frederick Goldin, pp. 3-46. New York: W. W. Norton & Company, Inc., 1978.

[In the following excerpt, Goldin explains the roles of history, Christianity, and loyalty in understanding the world of The Song of Roland.]

In the year 777 the Saracen governor of Barcelona and Gerona, Sulaiman ibn Yaqzan ibn Al-Arabi, appeared before Charles, King of the Franks, to persuade him to bring his army into Spain. Al-Arabi had revolted against the authority of the Emir Abd al Rahman of Cordova (a rebel himself against the Abbassid caliphs), and he now made the following offer: if Charles came to his aid against the Emir, then Al-Arabi and his allies (among whom was the governor of Saragossa) would submit to the authority of the Franks. This meeting took place in Paderborn. Charles agreed.

The King led a column of his army into Spain in 778, took Pamplona, and arrived at Saragossa, where, according to plan, he was joined by a second column approaching from the east. From that moment on, things went wrong. Al-Arabi's supposed ally, the governor of Saragossa, kept the gates of the city closed and repudiated the agreement made in Paderborn. Charles besieged the city. For a month and a half the Frankish army lay beneath the walls of Saragossa. Then, at the end of July, Charles decided to give it up and return to France.

On the 15th of August, in the year 778, Charles's army was stretched out along the narrow defiles of the Pyrenees, bound for home. What happened on that day is told in a biography of Charles written some fifty years later, the Vita Karoli by Eginhard:¹

…on their journey home in that same pass through the Pyrenees, they had to suffer for a moment the treachery of the Basques [Wasconicam perfidiam]. It happened this way: as the army was proceeding, stretched out in a long thin column because of the narrowness of that defile, the Basques [Wascones] lay in ambush on top of a mountain—the place is thickly covered with woods and therefore well suited for such covert attacks; and they rushed down upon the end of the baggage train and upon those troops in the rear-guard who were protecting the main army ahead, forced them down to the bottom of the valley, engaged them in battle and killed them to the last man; then they looted the baggage, and protected by the gathering night they scattered in every direction with all the speed they had. In what took place the Basques were favored by the lightness of their arms and the terrain in which they fought; and the Franks were put thoroughly at a disadvantage by the great weight of their arms and the unevenness of the ground. In this battle were killed Eggihardus, seneschal of the royal table; Anshelmus, count of the palace; and Hruodlandus, prefect of the marches of Brittany, among many others.

This disaster is the historic kernel of The Song of Roland. The presence of the Frankish army in Spain, as well as one can judge from shreds of evidence pieced together from many sources, both Carolingian and Saracen, seems to have been the result of a power play by the great King, who apparently saw in the internal strife of the Spanish Saracens a chance to extend his realm. It ended badly for everyone (except the Wascones), and it was a bitter memory to the King for the rest of his life, judging from the fact that the full scope of the disaster—the destruction of the rear-guard and the plundering of the baggage train—was not disclosed in any Carolingian document until after Charles was dead. No crusading intent can be detected in this enterprise, though there were attempts both before the event (in the benediction of Pope Hadrian upon the departure of the army) and especially afterward, to give it such a coloring, as though Charles had entered Spain to protect the Christians from the cruel yoke of Saracen oppression—an oppression that in fact did not exist.²

This is the poor, bare, inglorious event, a thwarted enterprise ending in a painful loss. Between this sad date of August 15, 778, and the composition—probably between 1095 and 1100—of the poem we now possess in the Oxford manuscript, [The Oxford manuscript (Digby 23 in the Bodleian) is the oldest known extant version of The Song of Roland, written between 1125 and 1150 in Anglo-Norman French. Since the manuscript is a copy at least once removed from the archetype, its dialect tells us little about the French of the original. For a study of the entire manuscript tradition, see Cesare Segre, La Tradizione della ‘Chanson de Roland’ (Milan and Naples, 1974).] there is an interval of some 300 years. The poem before us now has retained little of the historic event apart from the last name mentioned among the fallen and the annihilation of the rear-guard on the homeward march through the Pyrenees. Somehow this non-event has been enlarged into a great epic of treachery and loyalty, and this humiliating defeat at the hands of unknown brigands transformed into a holy crusade, a glorious martyrdom, a great apocalyptic victory ordained by God.

It is no wonder that nearly everyone who has studied this poem since the manuscript was rediscovered and then published in 1837 has been fascinated by the question of its origin. How did those unedifying events of 777-78 lead to the creation of The Song of Roland? Was it the work of a single poet or of generations of poets? Was it created all at once or by continual accretion over a long period? This much is certain: there are documents that show that before the composition of the Oxford Roland, during those three centuries, there developed an oral tradition centered on the battle at Rencesvals.³ “Someone invented the Emir Baligant, as someone invented Turpin's participation in the battle, as someone invented Oliver, as someone invented Ganelon, as someone invented the beautiful Aude.”⁴ Whoever put the Oxford Roland together might have invented Baligant, but he inherited all of the other important characters and events in the story, for they were famous before he did his work. Who he was, however, and how he worked, and what he had to work with, are all matters of dispute.

Whatever the circumstances in which it was composed, The Song of Roland looks back: it tells a tale that is set in the past. By the time of the Oxford version it was the remote and therefore the glorious and exemplary past, a golden age in the age of grace. It was looked upon as the time when the great dream of Christendom had come true, when a worldwide Christian community was established under a pious and crusading Emperor, and all men were bound in ascending loyalty to each other and to the Lord of all. The Carolingian Empire was seen as the fulfillment of a divine intention.

The numerous “errors” in this representation and in the figure and career of Charlemagne are frequently pointed out and sometimes rashly attributed to the poet's (or poets') ignorance. Charlemagne was in fact thirty-seven and not yet Emperor when the rearguard was ambushed: he was the King of the Franks, and he was not 200 years old; he wore a mustache but never a beard (which became à la mode in the eleventh century). Islam is and was monotheistic and forbade graven images of God: no Saracen ever prayed to the idols of that motley trio Mahum, Apollin, and Tervagant. The poet's many “errors” in geography have also been duly noted—Saragossa lies in the valley of the Ebro and not on a high mountain; and all those unidentifiable place names—though often at the cost of obscuring his poetic truth: for the fantastic name of every place locates it unerringly in the realm of God's enemies or in the sweet land of His servants. If one reads the poem as a chronicle, one can compile a tremendous list of such errors. But that is not how the poem should be read. The poet himself calls upon a chronicle, the Geste Francor, at certain times, usually to support what looks like a verifiable statement of fact, and in referring to such a chronicle he explicitly distinguishes his own narrative from one.

Judged rightly, these are not errors at all but essential elements in the picture that the poem presents. The great value one derives from studying the facts and identifying the “errors” is that one learns what has been rejected as unfit for the representation set forth in The Song of Roland. As it happens, nearly every detail of the historical incident has been rejected: history knew only a terrible defeat; the song reveals a glorious victory. But if Saragossa, that last pagan citadel, had really been set upon a towering eminence, or if Charles had really been 200 years old and at all similar to that patriarchal figure with the great beard white as the flowers of April, it is safe to say that the poem would have preserved, with lingering accuracy, these revealing historical facts; and we can be sure that the victory over Baligant would never have been invented if anything like it had ever occurred. For the poem sets forth the vision of an exemplary past—the past as it had to be, given the way things are; the past that guides the present and enjoins the future.

We see in the Charlemagne of the epic, not the historical king and emperor, but the true and accurate representation of an ideal ardently praised at the time the poem was cast into its present form, around the year 1100.⁵ Today we read a poem created in the distant past in which the poet looks back to a past even more remote, holding out to his audience the picture of an age when things were as they ought to be and all men were in their right places—an inspiring age that must be brought to earth again, when all Christian powers oriented themselves in homage to this great man, wise with the wisdom of 200 years of God's grace.

The positive historical context in which this vision of the past arose is usually identified as the long struggle of the Capetian rulers to centralize political power in France.⁶ From the death of Charlemagne in 814, the Carolingian kings presided over less and less, apart from the expanding dissolution of the empire; their lands and their powers were lost to the great barons, and in the end the king was a mere figurehead whose realm barely extended beyond Paris. Then, in 987, the last Carolingian king was gone, and a new line was established by Hugh Capet. From that moment on, he and all his descendants—including Philip I (1060-1108), who reigned when the Oxford Roland was composed—were involved in the struggle to bring the barons under the king's power. The Song of Roland is often held to be a kind of propaganda, a defense and glorification of royal power—the principle of the supreme sovereignty of the king is explicitly introduced into the poem in the trial of Ganelon, and it is authenticated in the judicial battle of Tierri and Pinabel, in which the judgment of God is revealed.

It is therefore a mistake to look in this poem for an account of the life of a particular period, whether in the eighth or the eleventh century: that approach leads to the preoccupation with “errors.” The poem shows us something else, something that a work of literature can show better than any chronicle or history: we see how an age regarded its past, recreated history in order to find precedence and dignity for its own aspirations. It may be true that the early Capetian rulers wanted some historical sanction to strengthen the poor foundations of their rule and commanded this version of the poem so that the glory of Charlemagne would shine on them. Whatever practical intention there may have been, the poem transcends it. The Song of Roland speaks to any age that wants to see its present struggles as something more than the madness of accidents and lusts, as something noble and necessary, as an ordained part of a vast integrity. There is a great difference, however, between the way in which we regard the past that engendered us and the way the poem looks to the past in which its action is set. The past it revealed to its earliest audiences was really a vision of their future.⁷ Those who shared that past were to give their support to the King's great struggle, a struggle that aimed not to progress from that auspicious time when angels came down from heaven and the sun stood still to help the Emperor defend all Christendom, but to return to it, to regain what had been lost: a perfect state pleasing to God.

Once this line into the past is begun, it can be extended infinitely further back. When Charlemagne, Roland, and Turpin pray, they look back to those who lived in an even earlier time but still command a vivid presence. Daniel in the lion's den, the three Hebrew children in the fiery furnace, Jonah in the belly of the great fish: these are not only the ones who came first, they are also brother heirs to Roland and Charlemagne and Turpin, sharers in God's loyalty and love, less separated by time from those who recall them in prayer than joined to them as fellow servants of the one Lord who protects them.

Therefore, in the figure of Charlemagne, religious and political aspirations are united; he is at the center of history's pattern, the nexus between the past and the future. For the same line that connects this old poem to an even older time, and to the beginning of time, can be extended into the future as well, to the end of time, when the empire of Charlemagne is upon the earth again. Those last days are foreshadowed by the action in the poem. For near the end of time, as God's word in Revelation foretells, the Antichrist will come who will suborn the human race when it is on the verge of sanctity, and he will rule until he is defeated by the angelic hordes of heaven—a battle prefigured in the song by the Baligant episode, whose apocalyptic imagery, as has long been recognized, shows the full range of the poet's vision.⁸

The poem's frame of reference, therefore, is infinite, ranging from before the beginning to after the end of time. And once we see the vastness of that frame, we can see how much is left out of it: we are left out, or at least everything personal and unique about us, the selves we know. For the vision of the poem is blind to our own linear notion of time: it cannot see that values change, it cannot conceive that one day human beings may stand in a different relation to those who rule and even to the Lord of all—to that Lord whose immanent justice and, in fact, whose very existence, they may, with perfect integrity, deny. In its apocalyptic vision and in its blindness to the conditions in which we make our way, it demands that we join in the struggle to bring back again the state and the age over which Charlemagne ruled. For it says in its very first line and repeatedly thereafter: nostre emperere! “Our Emperor!” We do not have to answer the poem's demand in order to read it aright. But we do have to see that it makes a demand, that the tale it tells contains an injunction upon the audience, that its vision surpasses the frame of its narrative.

When the poem looks to the future, then, it does not see us—at least not in the skins that we inhabit. It sees a pattern that involves all of the generations of mankind: the drama of redemption, foreseen by God. It is by the authority of that pattern that the poem enlists all of its audiences into the struggle to restore the state that God has blessed. We do not share that vision, and yet, because the language and style of the poem assume the immediate and corroborating presence of an audience—our Emperor!—even we, at this great distance, have a certain role to play. We are cast into that role by a world-view that is as deep and complicated regarding time as it is narrow and indifferent regarding space. This sense of organic time—time that stretches in a providential pattern over all creatures past, present, and future, and makes them contemporaries—puts us within the moral grasp of the poem and into its very action, no matter who we are and no matter how thoroughly we reject its feudal notions of faith and authority. For though we may reject that pattern as historical truth, we can still feel its effect as a poetic strategy, a way of calling upon the audience to complete the meaning of the song, and of heightening our awareness of the poet's conception of time and of the poem's origin in history. “Let no bad songs be sung about us,” says Roland, meaning: let songs be sung in our praise—and that means: this song, the Oxford Roland, witnessed by this audience—by us.⁹ The song's enduring ability to win an audience is just what Roland wished for, the fulfillment of all his boasts and promises, like his triumphant death. We today still witness the story of a brave man keeping faith, even though his faith is not ours, and our notion of bravery is, thank God, far different from his. For the song, by assuming our presence, puts us into the position of witnesses. We are, with all our disbelief, maneuvered by the providential strategy of the poem. As we read, we become that laudatory future now, and our attendance on the song is Roland's continual victory. We will most probably come away from this dialogue across the ages with all our disbelief intact; but as soon as we see how the song positions us, we can also see how the song presents itself, how it wants its narrative to be read, or heard: as a vision of history stretching across time and encompassing us, as revelation.

The Song of Roland is a chanson de geste, an Old French epic poem about the exploits (Latin gesta) of a great vassal in the service of his lord (or, as in certain later poems, in revolt against his lord). The lord that Roland serves is depicted as the Emperor of Christendom; Charlemagne, in turn, is in the service of the supreme Lord of heaven, and so the feudal pyramid rises above the world to end in the Author of all existence. The close relation between the epic genre of this poem, the feudal society it depicts, and the religious war that comprises nearly all of its action is the principle of its unity, and many errors of interpretation occur when one forgets what holds the poem together.

Paien unt tort e chrestïens unt dreit, says Roland, rallying his men (line 1015); Nos avum dreit, mais cist glutun unt tort, he says again, in the midst of the battle: pagans are wrong and Christians are right; we are right and these swine are wrong. Nowadays, of course, nobody has the right to talk like that, and so this famous exhortation is often condemned as a soldier's mindless partisanship. But in fact Roland is stating a major theme of the poem: the life of the feudal vassal can have no value unless it is sanctified by service to God. The pagan vassals are exact doubles of Christian vassals—they are brave, meticulously hierarchized, faithful to their lords; they wear the same armor, they have their councils, their battle-cries, their twelve peers, their famous swords, and their men of wisdom—and so the one radical difference between the two sides in this poem is exactly what Roland says it is, the fact that Christans are right and pagans are wrong.¹⁰ The pagans have devoted all of their virtues and their vast feudality to the worship of false gods; and so the greater their nobility, the greater their crimes and treasons. The pagans are loyal, but their loyalty is obstinacy, because they are against God and steadfast in their refusal to worship Him. The Christians are savage in battle, but their savagery is sanctified, transformed into the zeal of martyrdom, because they are justified by God. The poet goes to great pains to show how the Saracen structure reflects the Christian at every point; it is because they are the enemies of God and worship Mahumet that the pagans can never be more than reflections. Roland's famous utterance therefore means exactly the opposite of what it is often taken to mean. It is the warrior's expression of humility, his understanding that without the belief in God we are all glutun. Roland is a Christian vassal and knows that without the grace of God his great qualities would lead him to perdition.

One must always be mindful of the Christian inspiration of The Song of Roland. The immanent justice of God is the ground upon which the entire poem is constructed.¹¹ Every formal conflict in the poem is defined as a judicial battle whose outcome is God's verdict for the victor and against the vanquished. The pagans vastly outnumber the 20,000 Christians of the rear-guard; Baligant's hordes swarm over the countryside, far superior in size to Charles's army (ten divisions versus three times ten); the physically mediocre Tierri is by any normal standard no match for the great fighter Pinabel, whose towering presence intimidates all the barons of Charlemagne's empire—except for this one unremarkable man. In each case the miraculous victory of the smaller side reveals the will of God, for only He could have caused the astonishing outcome. Whatever propagandizing purpose it may have been meant to serve, the Oxford Roland is a religious poem because every event in it (including the establishment of the King's supremacy) is finally revealed as ordained by God, though there may be a secular precedent that looks like a cause; because all of its ethical values, though they are expressed in feudal terms, are ultimately justified as forms of Christian virtues; because all of its battles are cast as questions addressed to Him, and all of its victories are cast as His answer.

The ultimate religious reference of the poem clarifies its genre as well. To see this, we have to consider the kind of questions that the epic, especially the chanson de geste, is meant to answer. Many readers of The Song of Roland, for example, are distressed when they try to account for the motives behind certain actions, or to judge their moral value. What lies behind the hostility of Roland and Ganelon? Why does Ganelon explode with such murderous rage when he is named to carry out a plan that he himself had argued for? Why does he threaten Roland before the Emperor confirms the barons' choice? Above all, why does Ganelon commit treason? Or: Roland refuses to sound the olifant and thus ensures the martyrdom of the rear-guard. Should he have refused, many ask; was he right, or was he the victim of démesure, a term used by many who have written on this poem (but never by the one who composed it) to designate recklessness and inordinate pride?¹² Does he regret the deaths he might have prevented; does he repent before he dies? And soon we find ourselves plagued by questions that arise from the moral uncertainty of everyday life—but not from the poem.

For the poem tells a story purportedly based on history—and in history, too, motives are often obscure and usually held to be less important than events. Furthermore, though the poem is not unconcerned with human motives, it does not—cannot—recognize the values upon which our judgments of our contemporaries are based; and the rash condemnation of Roland's act often expresses nothing more than what we all would think of a man who did the same thing today. Questions such as whether Roland should have acted as he did are out of court. They address themselves to a false issue and raise doubts concerning values regarded in the work itself as beyond questioning; they are alien to the epic world, obscurantist, and anachronistic, for the alternatives implied in such questions belong to another time and to another genre, with another truth to tell. That other genre, in the medieval period, is the romance, a courtly form adapted to the depiction of individual moral experience. Some of the differences between these two kinds of narrative are spelled out in the following passage:¹³

The chanson [de geste] presents a coherent relation between one event and another, the romance a series of episodes that are, as far as their content is concerned, completely independent of one another. For the chanson, therefore, the surface development of the action is far more important than it is for the romance. … In the beginning there is an event that sets the action in motion; the action then runs its course almost on its own, autonomously, and the poem closes at that point where the chain of events (which began with that initial action) comes to an end. Every single episode has its fixed place in the totality of the action; it cannot be arbitrarily omitted or moved to another place without disturbing the progress of the narrative. The aesthetic unity that develops in this way is first of all a unity of the surface: it lies in the coherence and self-containment of the sequence of events. …

In the romance, on the other hand, each episode appears to be governed by coincidence and arbitrariness. … Its aesthetic unity is to be grasped not in the surface action but in the agent, the hero.

In the chanson, too, the hero can play a dominant rôle, but he is always in the service of the action, as its agent. One can see this most clearly in the fact that the hero can change in the course of the poem. In the Chanson de Roland, for example, the action does not end with Roland's death: from the point of view of the whole work, the most crucial part has not yet even come. After Roland, Charles takes over the action and leads it to a conclusion. This changing of the agent is not felt as a break, because the unity of the chanson lies not in the person of the hero but in the encompassing continuity of the action.

The motives of the epic hero are therefore determined by the action. They conform to a well-known course of events, which the audience is willing to accept as historical fact. The story related in an epic is always a version of some famous and significant historical event, some episode that is crucial, or held to be crucial, in the history of the people among whom the epic arises. Epic action thus has an historical core: it is centered on a past event. Even if the story is fictitious, the events in it are treated as though they had really occurred in the past and engendered the present state of things. For the epic is concerned with “actions irremediably completed.”¹⁴

Since the epic must have an historical core, it must always look back, and its point of view will always be anchored in the presence of its audience. Both the poet and the audience look back to a critical moment in history, the event from which their world emerged, the past they share in every performance. For the epic genre is rooted in the idea of performance; every epic poem needs listeners in attendance. Even if those listeners are utterly alien in time and place, they are bound to the poem. For if the pastness of epic action is essential, there must be a present audience—even if it is an implicit audience—by which that pastness is established and the continuance of its effect realized. Nowadays the old epic poem has an audience of readers, and we look back to other moments in the past to find the roots of our being. But the poem will not modify its demand on us; it insists with even greater conviction: Charlemagne is our Emperor. And so the poem's effect on us depends on our willingness to respond to its words, at least for the time that we play the role of audience, as though we were not alien to this remote part of the human past. For, in fact, this moment too is at the root of our being.

In this one respect we today can identify ourselves with those who heard this poem in the beginning, because we take up a similar position regarding the time in which the action is set. We corroborate its retrospective point of view, we complete its context: with every performance, or with every reading, the presence of the audience establishes the pastness of the action. The audience therefore plays an indispensable role in the creation of the epic world: it calls that world forth by looking back upon it. Now there is a famous passage in Boethius that analyzes what takes place when one regards an action in the past, and the terms he uses will be of great help to us. In Book Five the lady Philosophy speaks as follows:

And I will answer you by saying that a future event, when it is referred to God's omniscience, appears necessary; but when that same future event is considered in its own nature, it appears to be completely free and undetermined. For there are two kinds of necessity: one is simple, absolute—for example, it is by necessity that all human beings are mortal; the other exists only in a particular condition, so that if you know that someone is walking, he must necessarily be walking; for when something is known, it cannot be otherwise than as it is known. But this condition in no way involves the other kind of necessity, simple necessity; for what makes the thing that is known necessary is not its own nature but the condition that is added: no inherent necessity forces a man walking of his own free will to step forward; however, if he walks, he is necessarily stepping forward. In the same way, an event that Providence sees as occurring must necessarily be, even though it may have no necessity by its own nature. Now God sees the occurrence of future things that come forth from the freedom of the will, sees them as present things; and so these future things, if they are referred to God's vision of them, are made necessary by the fact that He knows they are to occur, by the condition of divine knowledge; but considered in themselves they do not relinquish the absolute freedom of their own natures. Without doubt, therefore, all those things will occur that God knows are going to occur. But some of these things happen as the result of free will; and these things, though they occur, do not by coming into existence lose their own nature, by which, before they occurred, they were able not to occur.

(V, 6, 124-126)

The world of The Song of Roland is ruled by conditional necessity, necessitas condicionis. The condition that makes things necessary is “added,” as Boethius says, to every event in that world, and it is we, the audience, who add it. Before we become an audience, we know, or are expected to know, what has occurred in history, or in the revered legend that the poem regards as the most authentic history; and so on becoming an audience we know what is going to happen in the tale we listen to: our ordinary knowledge becomes foreknowledge, becomes an analogy of divine providence, and every event in the epic is made necessary (in the Boethian sense) by the fact that we know it is going to occur, by the condition of the audience's knowledge. We know what happened in history; and since this or that event lies entirely in the past, we know further that it is a part of the history of the world thus far, that it participates in a transcendental design: it therefore has to happen, it is necessary.

Epic action is set in the past, as required by the genre. The historical event is now viewed within the frame of a narrative; it is integrated into a vast structure, endowed with significance, exempted from chance: the aesthetic form of the epic reflects the transcendental design of history. And since the epic event is thus completely enacted and framed, it is fully present to us in all its moments, we know it from beginning to end. Therefore, we can say, prompted by Philosophy: we know that we will walk, and therefore he must walk; for we knew, before the first notes of the song were intoned and we became an audience, that he did walk. We know—and our knowledge precedes every event, every cause, every motive—that Roland will refuse to sound the olifant: therefore, his refusal is necessary, for it is accomplished. The witnessed and believed authenticity of his act is all that counts. His motives, whatever they are, are at best secondary causes, completely determined by the action and significant in this poem, not because of what they bring about, but because of what they reveal: the loyal spirit of a true hero. We must regard his great spirit, his proud motives, and his famous act as praiseworthy, exemplary, pleasing to God, because they are necessary, foreseen, exactly as they occurred, in the destiny of Sweet France. For otherwise the Geste Francor would be nothing but the history of accidents and whims.

We have no right to ask: should he have done what he did? shouldn't he have considered another way? shouldn't he have been more reasonable? Questions like these deny the terribly necessity of history and the monumental dignity of the epic. They are questions brought in from the Age of the Team. Once we view the epic world from a true perspective, that of an audience witnessing the reenactment of an unalterable past, we see a world governed only by providential force. Within that world, however, considered only in his nature, the epic hero moves in his own present with undiminished freedom of the will; “I would be a fool to sound the olifant,” says he; or “I shall strike a thousand seven hundred blows.” The unpredictable present and the immutable past thus wonderfully coincide in an epic poem.

It is the past, or rather the audience's sense of the past, that ennobles these figures and their deeds, as it determines the form and technique of the poem. Because of the double perspective from which we, the audience, view the action—we see it looking back from our present, and looking forward from the hero's present—we experience at once in every figure and event the two forces of a free human will and a transcendental historic purpose. This is obviously true of an epic poem like The Song of Roland, in which this purpose is revealed as divine Providence: in Charlemagne's dreams,¹⁵ for example, or in the three judicial battles, the will of God announces itself. But even when all the figures and events are the consequences of a mindless causality devoid of purpose, we still recognize the presence of a transcendent design. That is because every epic presents its narrative as history, no matter if it is really a fiction. It demands of us that we regard its action, down to the rightness of the last detail, as a crucial part of the real past—of the living past, for the world that surrounds us as we listen derives from it. Even if some event were in its inception completely accidental, it was nevertheless caused, and it produced consequences that led, in turn, to a ramifying pattern of causes; and so, what began as an accident becomes bound by causality and consequence to the ineradicable continuity of the past. For if we were to recreate the fabric of history, we would need this event to weave into our pattern. No matter how it happened, it has led somehow to the present state of things, to the facts we find in our world and the condition of our community, and so becomes, as we look back upon it, a part of the providential past: as things have turned out, it has served a purpose; therefore, it is necessary.

In case all this sounds too theoretical and abstract, let us consider two concrete examples. We can get some idea of how a thing is ennobled by our sense of its pastness if we consider how the death of someone we have loved or admired affects our feelings about his life. Sometimes death is a frightening specter, when it is almost in our ken: we get glimmerings of it, especially when we are ill. At such times we are frightened because we suddenly become aware that our body follows laws that pre-existed us and will go on after us, laws we never made or even properly understood; all of a sudden our body no longer belongs to us, it is no longer the agent of our will, it responds to something that eludes our knowledge. Its governance is now taken over by another reality, another life, something that seems utterly alien to us but awes us by its immensity, its absoluteness, its intimacy. And so it sometimes happens that when someone we love has died, we identify his personal reality with that larger reality that ruled his body, and we find ourselves thinking of his death as an incredible achievement, for we magnify his image with the greatness of that which engulfed him. Somehow, in dying, he has grown larger, become ancient and infinite; and our sense of that vast causality blots out our awareness of chance and circumstance—his whole life takes on the coherence and inevitability of the laws that ordained his extinction. Everything that was gratuitious and accidental in his life now becomes inevitable with his death. For no law decreed his existence: he did not have to live; but once he lived, he had to die. What had seemed to be the purposeless welter of his experience—the meaningless color of his eyes, the incoherence of his enthusiasms, the unpredictability of his indifferences, the chance occurrence of his neighborhoods, the wild inconsequence that followed his choices—now, in the light of his extinction, reveals an inspiring necessity. Now it can be seen that all his sleeping and waking moments were required exactly as they were to complete the form of his life. This is when we think of him in a few essential poses—working with the tools of his trade, sleeping on his side with his hand under his head, listening to the news on the radio; and our every image of him is the illustration of an epithet—he is the skilled, the childish, the gentle one, the unjudging, the long-sleeping, the incorruptible—for we see him now transfigured with significance, a reality defined for all eternity. The things about him that irritated us when he was alive, if we remember them at all, we think of now as derived from his essential meaning. We regard him now as a perfect being, an aesthetic triumph, the fulfillment of an idea—a father, a prophet, a fool, an irreplaceably ordinary man, but in any case unique, for his specific life and death were ordained.

What about the death of someone we neither love nor admire? Here we can find an example in American history, in an event whose centennial coincided with the bicentennial celebration of our origins.¹⁶ Again we can see how conditional necessity transfigures men and events. In 1876 General George Armstrong Custer led the Seventh U.S. Cavalry, despite numerous warnings and apparently acting in disobedience of orders, into an ambush. He and his entire division were annihilated in the combined attack of the Sioux and Cheyenne, led by Sitting Bull and Crazy Horse. Since then historians and American history buffs have reexamined the various possibilities of the situation. Should he have attacked the numerically superior Indians? How would it all have turned out had he ordered the region to be properly reconnoitered, had he not been so hungry for fame, had he not desired a promotion so keenly, had he awaited (as he was supposed to) the arrival of the division led by his superior, had he not—like Roland—longed for a glorious victory?

We have a right, and perhaps even a duty, to ask such questions, because we know that this episode was completely unnecessary: the General could have acted otherwise. The massacre was gratuitous, the result of pure coincidence—it could have not occurred. For this event took place only because a fool on the one side and two brilliant strategists on the other found themselves in the same place at the same time.

And it is now more than a hundred years old, it is part of our past. One may claim, with some pride, that the passage of time has not conferred upon this event the dignity of necessity. And yet time and the sense of the past have in fact done their work. This event is permanently recorded in our history as an act of madness, the result of a single man's grotesque conviction and ambitiousness and a whole nation's crusading spirit and racist zeal; and it has, enlarged in this fashion, taken on something of the quality of the poet's marveling portraits of Ganelon. Even in raw history Custer's madness has achieved a classic ingloriousness, even a kind of satanic dignity, as a point of reference for the nation that went on to Vietnam, as the American locus classicus of moral imbecility and self-destruction.

But suppose this event were to be related in an epic. Then there would be another kind of transformation: all these accidents would immediately become inevitabilities. For the condicio would be added: lacking necessity in themselves, they would become necessary by being viewed from the perspective of a transcendent vision—providence and posterity know that these things will happen. We embarrassed and suspicious descendants would now become the audience, conscious both of the pastness of the event and of the epic form; and in that role we would, by our vision, transform the event into an enactment of necessity. Then the General's recklessness would become exemplary courage, an essential virtue of the American hero; the senseless slaughter would become a blessed martyrdom, the fulfillment of a sacred covenant; and above all, the defeat would become the first moment of an ordained rebirth. All these transformations would take place because our vision had enclosed the event in an historical and aesthetic frame that signifies that everything within is foreseen and therefore necessary. Looking back as inhabitants of the world that has emerged since then, we would see not an isolated catastrophe but the most critical moment in a providential structure. The question whether the General also acted as he did because he wanted a promotion would now be eradicated by the force of conditional necessity and replaced by motives better suited to the grand design.

Was the “real” Roland—the “original” uncrusading Hruodlandus, prefect of the Breton march, whose name appears, perhaps added by a later hand, in a few of the manuscripts of Eginhard's biography of Charlemagne—was this Roland, like Custer, a reckless fool, unmindful of the welfare of his men? Judging from all that we know of the original incident, Hruodlandus never had the choice that confronts the hero of the epic; he had no chance to summon help, much less to decide whether he should. All that we can surmise is that he fought bravely in a hopeless situation. We also surmise that nothing more specific was known about him during the genesis of the poem. Whatever the circumstances in which this poem came into being, there were no hard facts that could prevent those who sang the song from magnifying Hruodlandus into Roland.

But we know too much about Custer to make him an exemplary figure; there are too many witnesses to his nasty egotism and his lack of self-control. We cannot, without criminally deluding ourselves, give him the character of an epic hero, one that is worthy of receiving the commandments of necessity. Let us therefore give thanks that no one we know of has attempted anything of this sort, apart from a few foolish lyrics of the time and some “epic” movies since then. But simply from toying with this idea we can see that the hero's motives can never truly explain what takes place in the poem: they are mirrors, rather than movers, of epic action. For the hero would be diminished and his deeds trivialized if (like Custer) he had nothing but his own reasons and he were not appointed by history.

As epic necessity imposes coherence on the past, so it bestows dignity on men and events, for it removes them from the vanity of a personal will and identifies them with a divine intention. Necessity eradicates accidentality and creates, with its tremendous retrospective power, a need for every moment. The hero's motives are exactly right, even when he appears to a lesser man (like Oliver) to be most willful and undisciplined, for they realize a purpose too great for ordinary men to understand. He desires, with all his character and vitality, to bring about the crucial facts of history. Roland does what he does because he must do it, because the event has already taken place, in our view, and he has no choice. He is the agent of an accomplished action, and we are privileged to witness the true hero's graceful conformity to the rule of necessity.

Because we know the history of these events, we see Roland's acts as part of a pattern; and though we may later force ourselves to change our minds, we first see every pattern as the product of a deliberate will. For the point of view that sees, or projects, an immense design where there is only the welter of blind causality is a religious point of view; and so, even when we reject the substance of religion, we adopt its view of history when we are the audience of an epic poem. Roland's acts are part of an historical pattern, and we perceive them as emanating from the Will that produces history. Because of the ennobling effect of necessity, because his actions are always sanctioned by the demands of that transcendent pattern that we, in our present role as audience, cannot dissociate from the movement of a higher will, the hero can never be denounced as vain, or proud, or lacking in wisdom; nor, it follows, can his enemy ever be dismissed as simply a scoundrel. We must never judge Roland's motives by our common freedom and our common sense, because they are purely epic motives, his personal resolve to bring about what has already been enacted. Thus it is through necessity that the epic hero realizes his greatness and his humility; for he is the agent of providence.

The poet has exploited the effect of the pastness of the action and has arranged the narrative so as to put Roland's eminence beyond doubt; at the very least, if we do not simply admire Roland's actions, we refrain from judging them. But we are meant to admire them. Roland is the only one who knows the right thing to do when the Christians are confronted with Marsilion's offer of peace. The others have every conceivable good reason for accepting that offer. For one thing, nobody in the council knows that it is a trick. And Marsilion does promise to become Christian: his conversion was the object of Charlemagne's expedition. Naimon and Ganelon speak ably from principles—the highest principles. Roland speaks from experience—Basile and Basan had been slain earlier through Marsilion's treachery on a similar occasion: if Marsilion tricked them then, it is likely that he is tricking them now. They must not give up all that they have achieved for a mere promise; the war that Charlemagne came to fight is not over. But the others do not listen, for they are reasonable, and tired.

The scene in Charlemagne's court bears comparison with a scene in another medieval poem in which men make a choice in a necessitarian world, Chaucer's Troilus and Criseyde. The characters in that work are involved in a tremendous providential upheaval, the translatio imperii, the movement of the center of universal authority from Greece to Rome. Now for this great event to take place, Troy must be destroyed and Aeneas must make his escape; and for Troy to be destroyed, it must take in the agent of its destruction, the traitor Antenor. All of the characters are of course in the dark concerning the great moment in the history of salvation about to occur, and everything hinges on their reaction to the Greek offer to give them back their warrior Antenor in exchange for Criseyde. Hector sets before them the simplest and most obvious dictate of their moral code: their worth and dignity lie in their refusal to trade women. But the others think it would be mad to keep a useless woman when they can get a great warrior back. They are free to make their choice, and they choose the course that leads to their annihilation. Thus, as the matter is arranged by the poet, the freedom of the human will is preserved even as the force of necessity prevails; for though the Trojans did not know about the vast designs of Providence, they knew enough, especially after Hector had told them, to make the right decision, the one that would have excluded the traitor and saved the city.

The situation is worked out in the great English poem to reveal the moral failing of the Trojan populace; that same situation is developed to a different end in The Song of Roland. Like Hector, Roland tells the barons all that they need to know in order to make the right decision. We do not see in this any sign of a moral failing among the barons—they are all men of principle worn out with fighting—and least of all any sign of Roland's démesure or bellicosity. What takes place is the revelation of Roland's privileged position in the world. He is right, and all the others are wrong—and that is all: this scene does not show that he is right for the wrong reasons, or that Naimon is wrong for the right ones—two impossible and inconceivable ideas in the Middle Ages; it does not even show that his natural and unaided judgment is better than theirs. It simply shows that he is graced with the right decision. But in the world that this poem celebrates, he cannot be right by accident: not only his decision but his entire attitude is right—his militant response to the pagans, his whole sense of what a Christian knight must do is nearest to what pleases God, for it comes from God.

The massacre at Rencesvals does not begin with Roland's refusal to sound the olifant: it begins right here, in “the council that went wrong,” as the poem says.¹⁷ And we are expected to respond to Roland's act at Rencesvals in the light of the grace bestowed upon him in this council, remembering always that he is right, that the perilous position of the rear-guard would never have come about if his voice in council had carried the day, and that after Roland's death Charles ends up doing exactly what Roland had urged; so that the fact of his having been right then confers an authority upon his actions now, at Rencesvals. For judge him as one will, if he was bellicose, impulsive, proud, and foolish in council (that is how Ganelon judges him), he is all of those things now; and he was right then—these qualities led him, completely unaware of Marsilion's plan, to make the right decision. Now the pattern repeats itself at Rencesvals, and Naimon's wisdom is replaced by Oliver's wisdom. We the audience, knew that Roland was right in council because Marsilion's perfidy had been enacted before us. He is doing now what he did then, following the dictates of the Christian vassal's calling with unquestioning faith, though now we cannot see the corroborative light.

Still, it may be there, and the possibility of its presence, the inappropriateness of our moral categories to Roland's act, the very fact that what he does eludes definitive judgment, all indicate that we are not supposed to judge. It is an error as well to think that Roland regrets his act and repents before he dies. If he were truly to repent, it would have to mean that his act—his refusal to summon help—was free of all necessity; in that case he could not be an epic hero, and he would become the proper victim of the audience's common sense. For the moment we think that he should have acted otherwise, we deny the force of necessity. If there is no necessity, there is no providence; and if there is no providence, then Roland's heroism becomes an aberration, a state of madness, a terrible display of the private will gone berserk. Only from one point of view—Roland's own—is his choice completely free. From the perspective of the poem's audience, he had to do what he did, for the battle of Rencesvals was now locked into the great design of history—the glorious history of France. From that perspective, the action he takes is necessary, pre-established; and Roland's greatness lies in his willingness to carry it out. We are moved by his warrior's rapture as he speaks of the duties of the vassal, not because we share his notion of loyalty, but because we see his notion of loyalty leading him to fulfill the role into which, from the audience's point of view, history has already cast him.

Roland explains why he refuses to summon help: he says that to do so would bring shame upon himself, upon his family, and upon France. He also gives the reasons why it would bring shame. He believes that it is his duty to fight alone: that is why the poem makes him twice speak the famous passage on the vassal's duty at this moment. He is also sure that he can win, and it would be shameful to summon help when he and his men can defeat the Saracens alone, for he had sworn that Charles would not lose a single man, not a horse or a mule, as the army made its way through the passes.¹⁸ The concerns that led him to refuse to summon help—honor, lineage, sweet France—are named and praised by Charles, later, in his lament for Roland. These are the lights by which Roland acts. But we, the audience, enlightened by necessity, can see more: if Roland calls for help and Charles returns, the battle will lose its judicial character.¹⁹ Only if there is a victory of the few against the many can the outcome of the battle reveal the will of God. If Roland is right, then God has bestowed upon his outlook a special grace. He is the agent of God's will, the supreme vassal, and God has sanctified his calling, endowed it with a mission. In this we can detect an historical resonance. When the Church, in the interests of the peace movement and the crusade, set about enlisting the warrior class into its service, it did not try to temper the ferocious instincts of these men but rather attempted to train their martial spirit, in all its savage pride, upon a new, universal, Christian goal.²⁰

This may be the best moment to warn against importing twentieth-century sanctities into this 900-year-old poem. The Saracens in The Song of Roland, as the fantastic names and the ludicrous creed attributed to them make clear, are the postulated enemies of God, which meant in the Middle Ages that they are the force opposed to all human values. The poem is not, therefore, a genocidal tirade against the civilization of Islam; the Saracens here are demonic reflections, human souls degraded into automatons because they are without God—without reverence and humility. The true faith they lack is the faith of love, the commitment to the human community. Charlemagne naturally looks back to the time of Daniel, his brother-heir; but, except for the fathers who lose their sons, there is no kinship and fidelity among the Saracen generations—they have no past to look back to. They are agents of discontinuity, mad votaries who trample the idols they worship underfoot, who disintegrate the structures they mimic. Compassion for the enemy would be treason to man's community.

But that does not mean that the human reality of the enemy is obliterated in this great poem. Sometimes vehement hatred of another signifies a greater human commitment than pity, for pity often means that one has given up hope for the other's chances and, amid copious tears, written him off. It is said of Baligant that if he had been Christian he would have been a great man, a remark that is often mistaken as an expression of blind intolerance. But it is really an explicit recognition, in the poem's early feudal idiom, of the secular equality of Christian and pagan, and of the infinitely redeemable humanity of the enemy. The pagans are condemned, not because they are without God, but because they are without God by their own choice.²¹ They have used their heroic powers to thwart human value, but to their dying breath they have the chance to be converted by love. That is, as things turn out, a privilege reserved for Bramimunde alone. The others, if they are not killed by the sword, are converted by it. The poem's outlook is thus irreconcilable with the celebration of pluralism characteristic of our enlightenment; but it also excludes the ultimate nightmare, our nightmare, the vision that we dread: the tribalism of the industrial age, which regards those on the far side of the world as alien in essence, morally strange, incapable of our humanity.

Each man in this feudal community finds his place in a hierarchical structure of loyalties that ends in Charlemagne, to whom all are bound, as he is bound to them in the obligation to protect them.²² Each of the barons holds a position with respect to his own men—the men he brought with him to battle—analogous to that of Charles with respect to the great barons and ultimately to the whole community. The men are bound to the lord, the lord to Charlemagne. The word “man,” hume, is a technical term here, designating the sworn vassal, one who has done homage to a lord in exchange for protection, nurture, and gifts. The relation that binds man and lord, and man and man, in this way is designated amur, love; so that when Ganelon, at the height of his rage, shouts at Roland: Jo ne vus aim nïent (306). “I do not love you,” that is no understated and ironical insult but the most terrible thing he could utter. It means: the bonds of loyalty are cut, we are enemies. It is a desfiance, a withdrawal of faith, a declaration of war, and, in the feudal age, it legalized revenge. This is a point on which Ganelon pins his life at his trial.

We can see here how the poem depicts Ganelon always as a conscientious follower of the law and of the oldest feudal values. One of the many instances of perfect feudal love (in the sense defined above) occurs in the scene in which Ganelon parts from his men to go on his mission to the Saracens. Here we see the deportment of true vassals and a true lord: they beg to accompany him, he refuses to endanger their lives recklessly and provides for the succession of his son and the peace of his realm in the event of his death. And here we see as well the true Ganelon, the essential Ganelon—the man who, in his whole-hearted obedience to the law, subverts its intention and works the destruction of his community. For the effect of his brave departure is to sow the seeds of discord and to endanger the life of Charles's greatest vassal (lines 342ff). Ganelon the traitor is a pure creature of convention: his every word and deed are preceded by a passage indicating how someone regards him and what someone expects of him, and everything he does or says confirms that description and fulfills those expectations.²³ Depending on who describes or observes him, he is a raging malcontent, a faithful vassal, a false counselor, a revered lord of ancient lineage and long service, a liar, a protector. Note the utter inconsequence of his bringing up Basile and Basan (line 330) after he had argued in favor of accepting Marsilion's offer of peace. In recalling their memory he is restating a point made earlier by Roland (lines 207-10). He has completely reversed himself. His successive states of character are bewildering, all the more striking in comparison with the consistency of his effect: he is continually programed and reprogramed, and invariably destructive. Only in the trial is he named and defined forever.

The fact that Ganelon, the traitor, is repeatedly depicted as the perfect lord and vassal is eloquent testimony to the secret, indwelling weakness of the structure over which Charles presides. Ganelon never breaks any rules: he sees himself, and rightly so, as an upholder of the most venerable law, the earliest bonds of human community. [With one exception: he lies to Charlemagne about the death of the Algalife (lines 681-91), a clear act of treason for which he deserves to die. And yet, nothing more is ever said about it. In his indictment of Ganelon, Charles never mentions it. The poet wants Ganelon's treason and the justice of his execution to be beyond question to the audience; but he does not want the trial to be affected by this evidence, which, if it were mentioned, would have made the rest of the trial and especially the judicial combat unnecessary. The poet wants the trial to hinge on a completely different issue.] In his first outburst against Roland, he denounces his stepson for putting a kinsman in danger and thus breaking an ancient and prefeudal bond. And yet, Ganelon is a traitor. In fact, it is by following the rules that Ganelon commits his treason. That is the unspeakable wonder of Ganelon: he betrays his land by conforming to its law.

He denies with his last breath that he committed treason and insists that what he did was legal. This argument is no courtroom trick. His sincerity and the authenticity of his claim are reflected in the truly admiring portrait that the poet inserts at this moment (3762-64); in the admiration of his peers; in the unwillingness of his judges to condemn him, intimidated as they are by Pinabel, who will fight for him in the trial by battle and whose immense strength reflects the strength of Ganelon's position; in his status as a noble lord of a great and ancient family. He is no outsider, no Sinon, but an authentic member of this society and a passionate believer in its original law. He claims that revenge was his right and that he had fulfilled all of the established prescriptions by which revenge is sanctioned. “I took vengeance,” he says; “I'll admit to no treason in that.” For the right to take revenge is a basic right in the feudal community ruled by Charles, the oldest sanction of justice. Now the trial must determine whether there is a higher right than vengeance.²⁴

In the grandeur, sincerity, and persuasiveness of Ganelon is reflected the weakness of the entire system, for Ganelon, as he sees it, did nothing more than what the system authorized. And he is right; nobody had ever imagined the things their covenant could sanction. It is a fact that Ganelon could never have committed his treason had he lacked what the system provided and his own honored place within it. From his native status and his adherence to the law arises his treason. Without due process, without the prescriptions of custom, without a communal heritage of ethics and rights (as, for example, the right to take revenge), without those conventions that preserve the life of a community, Ganelon would have lacked the means to betray his native land. He is therefore the arch-traitor, for through him the system betrays itself. Ganelon plays an essential role in this system: he is its traitor. He brings to pass the unsuspected consequences of its fundamental laws, endows it with a shadow. For his presence is as necessary as the shadow cast by a body: if a body exists, its shadow necessarily exists; without the shadow there can be no body. Without the traitor or the traitorous force, there can be no system. In Ganelon we see, not a man who for one reason or another—rage, disaffection, avarice—becomes a traitor, but a man who was a traitor from the very beginning, a traitor by necessity, whose destructiveness is uncaused because his essence precedes every cause. For what can be the cause of his treason? It is true that Roland provoked him; one may even believe Ganelon's assertion that Roland cheated him—and these would be sufficient causes if Ganelon were only an avenger. But he is something more—the trial proves it: he is a traitor, and no motive conceivable in the poem explains that. His uncaused act can accommodate every cause conceivable in the feudal world. The mystery of his treason is a sign of his elemental being.

Ganelon is the destructive element of every secular structure, the indwelling cause of its instability. In him we recognize the traitorous possibility of every institution. Custom can betray, because it can preserve and reinforce an evil. The law can betray, because it can show the criminal how to commit a crime and be acquitted. Even loyalty can betray, because the object of one's loyalty may obscure higher values: Ganelon keeps faith with his family and his ethical code and just for that reason cannot see the supreme good of the Emperor's mission or the rights of the community sworn to fulfilling it. Praise of one's comrades, the longing for peace, piety itself can betray.

Ganelon's treason reveals that the precious rights and customs of every community will tear it to pieces if they are not hierarchized. Some rights have precedence over others, and the right of the King in fulfilling his God-given role has precedence over all. This is a hard-won principle, for according to the poem, no conflict had brought it down to earth before this, and it is not articulated until the trial, at the very end, by one man—“not too tall, not too short”—a man whose eloquence lies in his lack of personal brilliance and his willingness to stake his life on that principle. And still no one sees the principle, save Charles and his man: the barons do not see it, they are blind and filled with panic before the grandeur of Ganelon's lineage and the strength of Pinabel. They are reasonable, moderate, forgiving, and therefore almost traitors themselves (line 3814), for they do not know what is at stake—they bear no scars, they cannot see, they did not witness the martyrdom at Rencesvals. It takes the hand of God in the battle between Charles's properly nondescript man and the mighty and beautiful Pinabel to reveal the truth: the right of the King precedes all others.

Through the crime of its native-born traitor, the system has discovered its essential weakness, and a better system has emerged, the system of Charlemagne's great Christian empire. The grave losses caused by Ganelon's treason are redeemed, transformed into the precious suffering of rebirth, the moment he is declared a traitor. For a new state is brought into being by the treason of Ganelon, which appears as a shadow-act of the great treason that inaugurated the salvation of the human race, and by the trial in which he is condemned. For the first time, the reality of dulce France is fully established in the poem: sweet France is what he betrayed. For the first time, France is more than the remote and undefined object of the Christian's love and the Saracens' hate. Here, in the trial of Ganelon, it is involved in the action in the way that the other characters are, both as sufferer and as agent. The trial that defines the identity of Ganelon affirms the existence of a coherent political being, a state, whose presence is now immediate and effective. And this dramatic appearance of the state within the frame of the poem's action comes about for this reason: when something can be betrayed, that is proof that it exists, that it is no longer a mass of disunited powers but a defined being capable of engendering obligations. It can be betrayed because it is real and has the right to demand loyalty. And when this being is the community of a people and its king, then the treason committed against them is proof that they have established the institutions that make up a state; for Ganelon has shown that these institutions are the necessary condition of treason. And further, when this being declares itself betrayed, that is proof that it has become aware of its identity, that an act of self-consciousness, a prise de conscience, has taken place. And finally, when the state condemns its traitor, it reveals not only its power as an entity but also the ethical basis of its claim to loyalty: the destruction of the state through treason would be an arch-crime because it would threaten the basis of all human community, of human existence itself. Thus in the same act the state declares its identity and asserts its right to exist.

This is the precious moment for which every other moment in the poem was a preparation. We can recall, for example, the “impotence” and “passivity” of the King when Ganelon names Roland to command the rear-guard. Charles's immobility stems from the fact that he is legally paralyzed: Roland has not been chosen for the rear-guard as Charles's vassal, and so Charles is not in a position, even though he is Roland's lord, to order him to stay or go. Roland is chosen by an assembly of the barons, who, in giving the King their counsel, are acting as representatives of the whole state and, in effect, exercising the authority of the state.²⁵ That is why the King's weakness at this moment is a preparation for the great trial at the end: the King cannot become strong until an act of treason is declared.

Charles's behavior in this scene is usually interpreted as a reflection of the king's historic position vis-à-vis the barons at the time of the poem's composition. There may be something to that idea, but the full effect of this scene is clear only at the end of the poem. If this scene reflects an historical situation, it does not do so directly, for the barons are not shown acting in their own interests against those of the King. They are acting on a state matter, and Roland is named to perform a service to all. That is why it is a tremendous moment when Tierri says to Charles: “Roland was acting in your service”: for the first time the idea is put forth that service to all is identical with service to Charles, and vice versa, that Charles embodies the interests and authority of the state.

Thus it is only after the trial that it becomes possible to define the meaning of “sweet France” and the principle of its unity. Before that, we get only fleeting glimpses of a certain spirit shared by those who fight in that land's name: the courteousness and companionship of Roland and his comrades, the universal love for Charles, the sumptuousness and concord of the Christian camp. But these impressions are obscured by others: the disagreements in council, the sudden revelations of Charles's weakness, the private feuds, the disintegrative allegiances to family and private interests. The French, when they think of their land, remember their wives and daughters, their fiefs and domains (lines 820f). Thus the expression France has a certain effect but no clear definition: Baligant and the other pagans speak with perfect ease of “sweet France.”

Before the trial, France was defined only by its outer enemies: the hostility of the Saracens proved that France was dulce, pleasing to God, for it was locked in combat with the enemies of God. Now, however, through the condemnation of its native-born traitor, France takes on a native character and reveals exactly what it is that pleases God: it is a state in which all men are bound in loyalty through their ultimate obligation to the King, a state whose unity and well-being derive from the subordination of all privileges, rights, and interests to the King chosen by God. At the end of the poem Charles receives a new mission as head of a new born state.

Before the treason, when Roland was still alive, the God-given principle of the supremacy of the King was not yet clear, despite his own exemplary loyalty to Charles. Because of Roland's eminence, the greatest reverence could still be given to horizontal obligations—to family loyalties and the demands of personal honor—until Ganelon revealed the disintegrating effect of these unordered rights. But now all is arranged as it is pleasing to God, in a vertical hierarchy that ends, on this earth, in the King.

In other words: after the trial, as the system expunges its essential traitor, it also makes an end of itself: we do not see emerging from the trial the system that had always existed only now in a healthier condition, but rather a different system, something new and nearer to God's intention. Modern readers may be satisfied with an interpretation of the poem as a justification and celebration of a political and religious order. But for the audience that shared its past and its vision of the future, there was more. The action of The Song of Roland foreshadows, as we have seen, the great apocalyptic battle and the end of time as prophesied in the Book of Revelation. That last great battle will also be caused by treason, and it will end in the Last Judgment, after which the world will be set right again. Now since Charles's battle against the pagans prefigures that final battle, it must also prefigure that last stage in the restoration of the world: the historical event reveals the eschatological truth. Viewed in this light The Song of Roland resembles what Dante calls “the allegory of the theologians,” setting forth real events—things that really happened—as a revelation of the last things to come. And when the action of the epic is finally completed, when the historical event has taken place, then the condition of the world is in fact closer to the final eschatological perfection that the event foreshadows.

A new state arises from the destruction wrought by treason. The traitor has used to evil purpose all that was good in his society and has thereby served the good: his treason made the King strong. The poem draws a parallel between the crimes of Ganelon and Judas: an immeasurable good arose from both. Therefore this too, this ultimate service of evil to the good, is foreseen. In John 6:70 Jesus says: Have not I chosen you twelve, and one of you is a devil?—and these words filled Saint Augustine with wonder:²⁶

He might have said: “I have chosen eleven”; is the devil chosen, is the devil among the elect? One speaks of the elect in praise: is that man elect through whom, contrary to his will and without his knowledge, a great good arises? Just as wicked men use God's good works for evil, so God uses the wicked works of men for good. … What is more evil than Judas? Among all those who followed the Master, among those twelve, it was to him that money was entrusted and the care of the poor. And then, ungrateful for this great privilege, for so great an honor, he took in money and did away with justice, this dead man betrayed life, turned as an enemy against Him whom he had followed as a disciple. This is the great evil of Judas; but the Lord used his evil for good. He suffered Himself to be betrayed, that He might redeem us: behold the evil of Judas turned into good! How many martyrs has Satan persecuted? And yet had Satan ceased his persecution, we would not this day be celebrating the glorious martyr's crown of Saint Laurentius.

Thus it is foreseen that the traitor must participate in every structure and be the necessary cause of human amelioration. Here again we see that the poem wants its narrative to be judged in the light of providence; and in that light Roland's death is a great victory, and the cause of great good for all—martyrdom and a place among the flowers of Paradise for those who fell, a new state for those who survive and inherit the earth. Of this new state we are permitted to see only the circumstances of its birth and in Tierri its characteristic man. Its essential traitor has not yet appeared: that belongs to another story. And so we are shown a view of human progress, the providential cycle of treason and rebirth, which will come to an end in the conflict between the last traitor, the Antichrist, and the hordes of heaven—the great battle prefigured in the episode of Baligant. Through this painful cycle a small advance has been won, the foundation of a new state, whose good lies in loyalty, or, in the feudal sense, love.

ON METER AND STYLE

Every reader of The Song of Roland is struck by its distinctive and unforgettable style: the flat declarativeness of its lines, the dizzying shifts in tense, the absence of an explicit connection between one statement and another, the wholesale repetition of many passages, the frequent restatement of the same idea and retelling of the same event, the thinness of its vocabulary (consisting of fewer than 1800 words) and the rare use of figurative language, the powerful conclusiveness of each laisse (each group of lines), which gives one the feeling that the narrative repeatedly comes to an end and then resumes. The style is a powerful instrument in the hands of the Roland poet, a man who works with silences as well as words. We can trace here only a fraction of what he achieves with his strong meter, his formulas, his few words.

Many of the stylistic features mentioned above reflect the poet's use of the techniques of oral poetry—poetry composed as it is being performed, as the poet sings or chants or recites it before a present audience. In the last fifty years a great deal of light has been shed on the technique of oral composition.²⁷ Here, in abbreviated and simplified form, are some of the main points that bear on the style of our poem.

The performer of oral poetry, who is usually if not always unlettered, does not try to memorize a number of lines already composed, a “text”: considering the enormous length of some oral epics, that would be beyond the capacity even of primitive memory. Instead, he uses his memory to retain a great hoard of formulas that have been developed, honed down to perfection, over many generations and that he has learned in the course of a long apprenticeship and professional life. He possesses as well in his trained and capacious memory a great keyboard of narrative elements—longer elements (sometimes called “themes”) and shorter elements (“motifs”) such as, for example, the scene of the council, the arming of the warrior, the preparation for a journey, the journey itself (often depicted as a quest, a bridal quest or a quest for treasure), the death of the warrior in battle. Depending upon the occasion, the audience, and the length of time at his disposal, the performing poet chooses among the vast range of themes that would be included in the full, unabridged version of his tale. He composes the narrative on the spot, line by line, guided in his choice of incident and formula by the theme; for in his mind each theme is already attached to a particular subgroup of motifs and formulas, and his final choice, in performance, is determined by the narrative and metrical pattern.²⁸ He does not—could not—compose each line word by word, nor could he choose in each line from the whole, enormous common stock of formulas, for he would be paralyzed in mid-recitation by the sheer quantity of his choices.

In the Chanson de Roland, for example, each line consists basically of ten syllables divided by a pause (or caesura) after the fourth syllable (occasionally after the sixth). A formula of four or six syllables will obviously fill out the first or second part of a line. But a formula may also have fewer syllables, in which case it must be so designed that it can be joined to another brief formula to fill out one part of the line (or hemistich). The formula ço dist li reis (“the king said this”) fills up the first part of the line; and whatever the semantic value of ço, it does fill out the meter. If the attribution of speech is to occur in the second part of the line, the formula must contain six syllables: dist li emperere Carles. To take another example: ki a or est gemmee (“which is set with precious stones in gold”) is a frequent and useful second-hemistich formula: it can be applied to different objects (sun elme, “his helmet,” line 3142; la bone sele, “the good saddle,” line 1373) and turned around to fit the assonance (ki est gemmee ad or, line 1587). It is an essential principle of oral poetry that whenever the same combination of metrical, lexical, semantic, and phonological conditions occurs, the same formula is used: this is the principle of economy or thrift. …

Is The Song of Roland an oral poem? Opinion is divided, and the entire question has often—too often—been reduced to a dispute between “traditionalists” (who believe in the continual re-creation of the poem by generations of anonymous poets) and “individualists” (who believe in the unique creation of the poem by a single poet, perhaps named Turold).²⁹ The extreme individualist position, as we have seen, is now untenable because there is evidence that before the composition of the Oxford Roland a legend had developed concerning Roland, his lord, his companions, his betrayer, and his enemies.

Anyone who argues the traditionalist position must insist that the Oxford Roland is an oral poem, for otherwise this position is equally untenable: the only kind of poem that develops by slow accretion through many generations is an oral poem, untouched by the definitiveness of a written “text,” a poem of continuous potentiality, brought forth anew from the great matrix of formulas and themes each time it is performed—a poem “that lives in variants.”³⁰ Traditionalism therefore regards the Oxford Roland as an accident: of the actual and potential variants, which are as infinite and impermanent as the patterns of a continually revolving kaleidoscope, one version (and not necessarily the greatest one ever performed) has chanced to get written down. For there can be no talk of “development” without the assumption of this fluidity. That is why traditionalism must consider the manuscript an anomaly, and it warns us continually against attributing any kind of scriptural authority to it as though it were the “author's” last word:³¹

… the manuscript is a pure accident: it represents nothing more in effect than the recording through writing of some one of the innumerable versions that do not cease being born as long as the song is sung. … The manuscript of an individual poet represents an enduring reality, that is, the text fixed by an author; whereas the manuscript of a traditional [i.e. oral] work represents a fleeting moment of a multiform reality.

There are, in fact, other versions of the story of Roland extant, and though these are later than the Oxford poem they are without question based on earlier material and may very well argue the existence of countless other renditions.³²

Now it is clear that the techniques of oral poetry are at work in the Oxford Roland, for part of it is composed by formula and theme. But that does not prove that it is an oral poem. And so the question remains: does the Oxford manuscript contain a copy of a unique creation prepared (not necessarily written) in advance of performance; or is it the accidental recording of one performance among a vast number that have been sung into the air? The question is of some importance, for it affects the way we read the poem. If we believe that it was prepared in advance, then we can read it as a “literary” work and respond to its words, images, characters, and events as we do when reading written poetry. We are free to look back, reread, compare part with part, consider what happens to a word or image each time it occurs, study the retroactive effect of one turn of events upon all that has preceded: we respond to the work as a construct designed so that every part affects every other. If we believe that it was composed in performance, then we read it as an oral poem, and we guard against the reader's habit of close reading. If we want to explain why a certain word or image or event is there, we must be careful to remember that it belongs to the theme and to the group of formulas that the theme has attracted: it is there because of the technique of oral composition. We have to block every exegetical impulse; the only context we can refer to is the context of the line—the meter, sound, and sense of ten syllables—and the theme. We must refer every element to a pre-existing, universal, unrestricted repertory of narrative units and fixed expressions, and to the rules governing their use: we must not explain it by some other element in the poem, especially one that is remote from the local context—that would be relapsing into our reader's habit.

These are the alternatives that face us if we take a hard line in this matter. When we come upon the word martyrie (“slaughter,” “martyrdom”), for example, should we refer to all the other passages in which that word appears and consider whether it takes on an added meaning or an ironic connotation as a result of its frequent recurrence? Or should we say that it is there because of the nature of formulaic poetry, that the poet uses the word each time it occurs, not because he wants it to reflect upon an earlier use of the word, but because it fits the meter, the sound pattern, and the narrative moment and because it is traditionally associated with a certain narrative theme in the field of warfare? If we believe that the poem was prepared in advance, we refer every element to a system within the poem; if we believe that it was composed in the pressure of performance, we refer every element to a system outside the poem. That is the hard line.

Obviously we can never know for sure whether this poem was prepared in advance or composed on the spot. There is, however, one thing that is sure: if we make the same demands of this poem that we would make of a written text, it responds with amazing beauty. Let us take that same word martyrie as an example. If we bear in mind that it can be used either in the sense of “slaughter” or in the spiritual sense familiar to us, then we can see that it is often used with wonderfully illuminating irony. In line 591 Ganelon warns Marsilion that he must pay a price for the attack on the rearguard: ne l di por ço, de voz iert la martirie: “Your men will be martyred.” In betraying the Christian side he still thinks in Christian terms, and he unintentionally reveals how the pagans are doomed to die a worse death than the death of the body: they will be “martyred,” but to what cause? In line 965 Margariz boasts of what he will do to the Christians: li. xii. per sunt remés en martirie: “The Twelve Peers await their martirie”; he means “slaughter,” but his words contain a truth—the promise held out by Archbishop Turpin of eternal honor among the flowers of Paradise—that Margariz cannot imagine. In line 1166, Roland says, cist paien vont grant martirie querant: “These pagans seek a great martirie.” Here one might argue that, in the context of battle, martirie simply means “slaughter,” without any religious connotations whatever; but this argument, far from disproving an ironic use of the word, explains what makes it possible. For if the word did not have a literal and obvious sense to begin with (“slaughter”), it could not have a second, or ironic, meaning (“martyrdom”). Roland's words mean that the pagans are coming to be slaughtered and that they will be “martyrs” to an unholy cause. The pagans' “martyrdom” is, like the pagans' feudal system and their apparent virtues—their courage and loyalty—an exact image of the Christian reality, only grounded in error and ending in perdition. In line 1922, on the other hand, Roland uses the word in both senses without any ironic intent: Ci recevrums martyrie.

Is it possible that the Chanson de Roland came into existence as a written text employing a number of oral formulas and themes? For a long time this possibility was denied: there was no such thing as a “transitional” poet, one who could write and still feel free to make use of the oral-poetic system. A poet was either “oral” or “literary” and would use only the techniques appropriate to his type. But now there is evidence that the Chanson de Roland (as well as other medieval epics) was composed precisely in conditions formerly considered impossible.³³ A great poet—“the last redactor”—using material that was preserved in oral tradition (and perhaps by other means as well) composed a text intended for oral delivery. And because this poem was meant to be performed, he used the techniques of oral composition: those were the techniques employed in the poetic material he inherited, as well as in most vernacular poetry, and they were adapted to the entertainment of an attending audience—just as the principles of rhetoric, which began as an oral art, have been preserved and venerated for centuries in written prose. Thus the poet was not bound by the principle of economy and could make use of a wider choice of expressions and techniques than those of purely oral poetry. If this is true, then we should read The Song of Roland as words on the page, a literary text.

Nor can we, in fact, read it in any other way. No matter what traditionalists and individualists may say, the alternatives we have been examining are not mutually exclusive. We say that if it is a literary poem, it is prepared in advance; but if it is an oral poem, how can it not be prepared in advance? The lettered poet can write several drafts and revise indefinitely. But what does the oral poet do? He retells a tale that he has heard and told many times before: every past performance is an earlier draft, and every present performance in effect a revision. The oral poet would, to be sure, find the idea of revision incomprehensible and would swear that he is retelling a story exactly as he has heard it; but we know that in fact he does not tell it the same way twice—the very nature of oral poetry rules that out, for that would be tantamount to memorizing a text. He may not intend to alter his narrative, but he does: The effect of his repeated performance of the narrative is to refine and improve his treatment of it.³⁴ Even if he does not think in terms of making the poem better, even if in most cases he only makes it different, all these distinctions fall when a great poet appears: the line between alteration and revision becomes very faint then, if it exists at all. He knows what he is going to say—the tale is mapped out by themes and the formulas that they attract—and he is free to use a certain word or image or episode now because he knows he will use it or something similar later: The Song of Roland is distinguished by this parallelism. The whole song is present to the poet. He therefore has a sense of context similar to the literate reader's. A strictly oral poet would not “revise” as a lettered poet does, for the one composes word by word, the other by groups of words. Nevertheless, what shall we call his refinement of the poem but “revision”? Besides, the “strictly oral poet” is a disputed postulate as far as this poem is concerned. Only about twenty-five percent of the lines of The Song of Roland contain word-groups identified beyond question as formulas. A good part of the poem may have been composed with greater freedom. As far as the question of how we should read this poem is concerned, the difference between the two kinds of poetic composition boils down to two issues: the poet's perception of what he was doing, and the conditions in which he worked on the earlier versions—two issues about which we can never have sufficient knowledge and which in any case have no bearing on our response to the poem before us.

Then let us respond to every feature of this poem and not fear that the procedures of its technique can invalidate our response. Every feature affects us in a certain way, and the capacity to produce that response is part of its essential reality. For example, the poem is effectively anonymous: the name Turoldus in the last line may be the name of the poet, or the scribe, or the redactor, or the performer, or the author of the source, and the mystery of his name only intensifies the effect produced by anonymity. Most epic poems from this period are, as it happens, anonymous; but if we rest content with that observation, we imply that the anonymity of The Song of Roland is accidental and therefore irrelevant to our response. But in fact it affects us profoundly, just as we are affected by the appropriate mystery of Turoldus' name; and if we define that effect, we shall discover something essential in the poem. For its anonymity reflects its style, and its moral disposition. In our response to the poem, its anonymity is a positive sign of its mode of expression, which is enormously powerful without being in the least idiosyncratic. In that same anonymity we find its ethical bias confirmed: its unconcern with subjective experience, its celebration of a communal heritage, its sense of the mission bestowed upon a people newly chosen in the age of grace, its injunctive force trained with visionary blindness upon all of its audiences—all who hear these words must struggle to restore the kingdom that once pleased God—a force that no man speaking in his own name could ever command.

We may react to this anonymity as we would to an ancient ritual gesture, if we regard it as something that was once ordained and significant, a revelation to a circle of witnesses, the poem's mark of its own authenticity. We are about to hear, not the imaginings of some vainly “original” poet, but a privileged vision of history, beheld long ago by a nation that believed itself called to the service of God. Its anonymity identifies it as something monumental and inherited, with an authority that transcends the credibility of any one person. It is therefore analogous to the anonymity of the Scriptures: it declares that the true author is greater than any man who can be named, the events it narrates are more fraught with meaning than any man can imagine, the truth it reveals is higher than any man can conceive: it is a monument, it belongs to all.

Something needs to be said in this connection about the role that formulas play in epic style, and particularly in the style of The Song of Roland. Because they are meant for limitless repetition, formulas do not strive to be brilliant or literary. The more “original” and unexpected any expression is, the less it can bear repetition. The formula has to be honed down, polished, made spare, indivisible, uniform, inconspicuous with regard to its literary merits, serviceable, purified of every element that may restrict it to a single use or a single context or even a single poem. The deathless line belongs to written poetry and usually to an individual talent; moreover, it is not a greater cause for celebration than the true formula, which will keep its force and deliver its message no matter how many times it is repeated. Milton's line “When Charlemain with all his peerage fell / by Fontarabbia” makes us pause in admiration, with that astonishing burst of sound at the end. But our high estimate of this line and of the poet's talent would be utterly wiped out if we found him using it another time, or every time the occasion could accommodate it. Par amur et par feid, “in love and loyalty,” does not arrest our attention with its brilliance, for the oral poet does not want us to miss the movement of his narrative as we ponder the beauty of his words. The formula gives us the pleasure of feeling a metrical demand fulfilled, the keystone fitting into the place reserved for it. Some of these formulas are certainly beautiful by any standard and would glisten in a work meant for reading; but it is not such beauty that the poet strives for above all—just as individual bricks and stones may have a special glint and texture, though their greatest effect lies in the part they play in the whole structure: whatever that part may be, they must not detach themselves.

It is not enough for a formula to have a certain metrical shape: it has to have as well an indestructible semantic content and ethical integrity. That is one of the reasons why an oral formula sometimes requires generations to perfect: it is an expression that has a certain metrical form and that does not become degraded with repetition. “Don't call us, we'll call you,” though it has a good six syllables, is the exact opposite of a formula.

The brilliance with which the Roland poet uses the formulaic style has long been celebrated. You hear the same formulas that were used to describe Christian victories now being used to describe pagan victories, and that is frightening (compare lines 1228 and 1576, 1233 and 1578, for example). But the great power of his style cannot be conveyed by such strategies, effective as they are.

We call it a formulaic style even though only about a quarter of the text has been identified beyond question as consisting of formulas. For one thing, it is reasonable to assume that many other expressions are formulas, even though we have not been able to find them used elsewhere. In any case, as far as the style is concerned, this is not a key issue. We may need to know what percent of the text consists of formulas in order to decide whether the Chanson de Roland was an oral poem, but we do not need this information in order to see that it is in the style of an oral poem. For the only way in which we can identify a word-group as a formula is to find it used elsewhere in the text or in other texts; a formula cannot be distinguished from another word-group in the poem by any other objective standard. If one compares hemistichs identified as formulas with those that are not, one finds no differences in meter, vocabulary, or phonology—a mun espiét trenchant (line 867) can be identified as a formula (compare lines 3051, 3114, 3351, and others); ki ben trenchet et taillet (1339) cannot, although each word in this group appears several times in the poem. If the Roland poet in fact made up wholly new lines, he drew from the same lexical field, he used the same kind of words and constructions that had gone into the making of the formulas.

No matter where he found his lines, the most important thing with respect to his style is that he composed in word-groups rather than single words. Every expression was designed to fit into a four-syllable or a six-syllable mold with a tonic accent at the end, and to be instantly recognized by the audience as part of a traditional motif: when a warrior boasts of what he will do to his enemy, or when he is actually engaged with his companions in bitter combat, something will be said about the sharpness of his spear or the cutting edge of his sword, something familiar and anticipated with pleasure. The technique of composition consisted in molding a grammatical unit—the subject well known to all (li quens Rollant), the predicate affirming a famous act (vait ferir le paien), the modifier fixing the physical and moral coordinates of the feudal world (a lei de chevaler), or any combination of these (cil sunt prudome)—into a pre-established metrical form.

The question that concerns us here, therefore, is not whether this or that half-line can be identified as a formula; but given the fact that the entire poem is composed in the style of the formulas, with its metrical demands and its feudal vocabulary, how shall we define the power of that style, and how does the poem make use of that power?

Formulaic language is the refined product of a long tradition and is therefore impersonal, objective, authoritative; it cannot be toyed with or used idiosyncratically. If the theme is war and the motif is the death of the warrior in single combat; and if, when the time comes for the warrior to fall or for his blood to flow, there is a need to define in the first hemistich the direction of this movement, then that need may be filled only with a formula, like sur l'erbe verte (“on the green grass”), or a formulaic expression, a word-group that sounds like a formula because its language and structure are the same—in any case not with the kind of surprising and inimitable phrase that we admire in later poetry and identify as the hallmark of a certain poet. In the old epic poem every word-group comes forth as an inevitable expression, both familiar and obligatory, and the property of all.

This was the style prescribed by tradition. And this language, bound as it was by strict rules inherited from the past, was the language of authority, the language reserved for the telling of the nation's history in song. Thus the formulaic style and the historical content of the chanson de geste—the great deeds performed by noble ancestors—were inseparable, even indistinguishable. That history was the precious heritage of the feudal community; and apart from charters and documents (like the Geste Francor to which the poem refers), it was only the song that preserved the memory of the past; only in its language was the wisdom and greatness of those who came first made available to those who lived now.³⁵ The words and constructions of the song became the language authorized by tradition for the recital of those great events that revealed the destiny of “sweet France.”

Now in using the techniques of oral poetry, Turoldus (if that was the poet's name) was a master of its injunctive and authoritative style. We can see the effect to which he used formulaic language—that is, language that consists of, or emulates, oral formulas—if we recall for a moment our earlier discussion on the nature of epic action. For the power of The Song of Roland lies in the relation between necessity and formulaic style.

The formula returns, obeying a law of its own, beyond the pathetic field of the hero's experience: Hector is “tamer of horses” by a rule to which his pains and joys are irrelevant. The blood of the dismembered Ganelon is li cler sanc, and it falls sur l'erbe verte, as the blood shed by many victims, both pagan and Christian, was bright and famous and flowed upon the green grass. The formula appears according to a transcendent system of versification, whose operations are never affected by the will or the condition of those involved in the process of the action. That system—that strong meter demanding a fixed and venerable expression—cannot take note of any unique or unprecedented experience: it is a system of inflexible word-groups, each of which defines a segment of a universal motif; every fallen warrior's blood flows alike—before the warrior falls, it is already determined how his blood will flow. Every formula occurs, and every newly coined formulaic word-group seems to occur, because the system demands it. That system is therefore one of the structures through which necessity exerts its force. Necessity operates in the appearance of the epithet—the epithet appears because it must, because it is, in the light of the system, foreseen.

Every great poet who uses the oral-formulaic technique makes the force of that system his own. As an apparent product of necessity, the epithet reflects, not the action that is going on at any particular moment, but, depending on the context, the glorious stature of the hero as the executor of a providential design, the indispensability of the traitor, the irreplaceable foppery of the king, the fateful compassion of the friend. Hector is “tamer of horses” even when he is being pursued and then transpierced and his body desecrated in the full sight of the family to which he was devoted. The epithet occurring now defines the true freedom of the heroic will—the freedom to chose what necessity decrees—and completely ignores the subjective experience; it continually recasts the sufferer in the role that fate has foreseen for him. His suffering is ordained by the same necessity that chose him as its agent and that now in another mode commands the occurrence of the formula. Now we see him ennobled by necessity: the epithet that celebrates his role transfigures his suffering, associates his pain with his heroic privilege, with the sanctions of a universal order, and prevents his dignity from being obliterated by the spectacle of his undoing.

So, too, the formulas that depict the death of Ganelon come forth, in the work of this great poet, as the result of necessity. The rule that prescribed the four-syllable formula and the six-syllable formula in the line: Sur l'erbe verte en espant li cler sanc (3972), corroborates the meaning of the next two lines—

Guenes est mort cume fel recreant.
Hom ki traïst altre, nen est dreiz qu'il s'en vant!
(Ganelon died like a traitor, like one who broke faith.
When one man betrays another, it is not right that he should live to boast of it!)

None of the hemistichs in these two lines can be positively identified as a formula (they do not occur elsewhere in the poem), but they are strikingly formulaic, in the sense that we have defined—the words fel and recreant and key words throughout the poem; and that last sententious line recalls line 3959: Ki hume traïst, sei ocit et altroi (“A traitor brings death, on himself and on others”). Ganelon's death fulfills an ancient and universal injunction: a traitor must be destroyed. Now we see that the execution of Ganelon has taken place by the ultimate authority of a design. The sequence that has led to this event was prescribed, like the long-established order of the words that commemorate it. For necessity operates everywhere: in the death of the traitor, in the singing of the song. The act of treason was necessary, the punishment of the traitor was necessary—the memory, the pattern of the song, the presence of the audience, the mission reaffirmed in the performance, all appear as though foreseen, and they are reflected in the language of necessity, the formulaic style.

We see this again in those remarkable moments when the phrase dulce France is uttered by a pagan—Baligant, for example, or Blancandrin. He says dulce France because the line demands this formula, but the rule that determines its appearance has all the inexorability of the law that condemns the speaker. In saying dulce France he confirms his exclusion from the land of the blessed, his willing enactment of the role assigned to him by providence. The inexorable return of the formula is a mode of necessity, patterned like the action that takes place in the world of the epic.

The demonic stature of Ganelon is revealed through the formulas by which the account of his death is recited. His life and death are foreseen; the evil he did is now located in a divine plan—for this religious resonance is inherent in the formulaic style. The word order of each formula and the rule that determines its use are fixed by the ancient customs of a close-knit people. Now all of these features—the immutable phrasings inherited from a remote past, the revered tradition that prescribes the times when these words must be uttered, the implicit attendance of an initiated audience united by a common heritage, the commemoration of a decisive act in a necessitarian world—are essential features of a sacred text.³⁶ The formulaic poet stands to the tradition governing his words in a relation formally identical with that of the scriba Dei to the divine source of his text; and his performance participates in the nature of a rite. He “makes up” nothing, he does not “choose” his words like the solitary prodigy admired in a later age, he does not own his text: he is the instrument—the authentic voice, the privileged hand—of a higher being and a greater will: the mystic source of community.

The poet may, as we have seen, invert the formula, adjust it, for the sake of the meter and the assonance, but his action is never capricious or arbitrary. It is prescribed, and it is therefore the sanctioned exercise of the words' power. The formulas he uses contain remnants of an earlier stage of his language—utterances once heard in the Land of Fathers, archaic words, old forms that preceded changes in sound and idiom, preserved intact by the immutability of the phrase.³⁷The Song of Roland contains many echoes of the old language; and this, too, the presence of archaic formulations hallowed by tradition, is characteristic of a sacred text. The Song of Roland was the monument of its audience and of all the audiences recalled in each performance, it was the secular scripture of their community.

Each laisse was a relic and an oracle, retracing the history and revealing the future of the audience—and of the song as well. It commemorated in its archaic phrases all of the past performances of the poem witnessed by the forebears of the audience, and it confirmed the promise that these same words, fixed as they were forever, would be heard by the descendants of those who now attended. Each laisse celebrated the past and the future of the nation and of the song; and so, when it was chanted in the ritual of performance, it reflected upon the nation the immunity to time and the integrative power of the song, and affirmed the communion of all those who listen, have listened, will listen. For the formulaic laisse immobilized time, transformed it into an edifice, a human structure infinite in its accommodation and ordained by providence, in which all the heirs of Charlemagne were assembled. Through the miracle of the singing of the song, their forebears became their companions, the empire of their descendants was restored, Charlmagne was nostre emperere.

Therefore, each laisse contains not so much the linear narration of a certain act as a meditation upon it, a reflection upon one event from several different points of view (the agent's, the audience's, the providence of conditional necessity) the points in time (the agent's present, the audience's present, the chronicled past, the revealed past, the political future, the apocalyptic future). This can be seen most readily in the laisses parallèles and laisses similaries—successive laisses rooted in the same moment in the narrative³⁸—but it is the controlling principle nearly everywhere. The action is rarely advanced in each of these narrative units; even in those laisses where there appears to be plenty of action in rigorous sequence, as, for example, in the case of “the epic blow,” the narrative is halted, rather than advanced. In lines 1642-51, for example, we follow the path of Roland's fearful blow through the helmet, the nose, the mouth, the teeth, the body, the silver saddle and the battle horse of his enemy. This sequence is described in the present tense. Then the action is chronicled, in the simple past, the doumentary tense: “He slew them both”; and the chronicle line is the only one that advances the narrative.³⁹ Otherwise, the action is refracted, broken down into its elements, stopped, recollected, and then set into the vast mosaic of the past. For the passages describing “the epic blow” are intended not to advance the action but to reveal the quality of the warrior who strikes.

The laisse is therefore a circumspection rather than a narrative, a reminiscence and an anticipation focused upon an action that is often (as, for example in lines 1902-3, when the tense shifts suddenly from present to present perfect) not even, strictly speaking, narrated. Each laisse is a lyrical arrest, and this static, meditative, reverential attitude of commemoration is one of the most distinctive features of The Song of Roland.

Notes

1. Translated by Lewis Thorpe in: Einhard and Notker the Stammerer, Two Lives of Charlemagne (Baltimore: Penguin Books, 1969). The translation in the Introduction is the present writer's.

2. For a full account of the incidents of 777-78, see Jules Horrent, “La Bataille des Pyrénées de 778,” Le moyen-âge, 78 (1972), 197-227. See also Martín de Riquer, Chansons (see Bibliography, below), pp. 13-21; and Paul Aebischer, Préhistoire et protohistoire du Roland d'Oxford (Berne, 1972), pp. 13-92.

3. For a survey of all the documentary evidence concerning the development of the story of Rencesvals, see the works listed below by Martin de Riquer and Jules Horrent; and Ramón Menéndez Pidal, La Chanson de Roland et la tradition épique des Francs, 2d ed., tr. I.-M. Cluzel (Paris, 1960).

4. de Riquer, Chansons, p. 88.

5. See Robert Folz, Le Souvenir et la légende de Charlemagne dans l'Empire germanique médiéval (Paris, 1950); Heinrich Hoffman, Karl der Grosse im Bilde der Geschichtsschreibung des frühen Mittelalters (Berlin, 1919).

6. See Erich Köhler, “Conseil des barons” und “Jugement des barons”: Epische Fatalität und Feudalrecht im altfranzösischen Rolandslied (Heidelberg, 1968); Barnaby C. Keeney, Judgement by Peers (Cambridge, Mass., 1949); Jean-François Lemarignier, Le gouvernement royal aux premiers temps capétiens (987-1108) (Paris, 1965). See also Karl-Heinz Bender, König und Vasall (Heidelberg, 1967).

7. See Matthias Waltz, Rodlandslied-Wilhelmslied-Alexiuslied, zur Struktur und geschichtlichen Bedeutung (Heidelberg, 1965), p. 24.

8. See Karl Heisig, “Die Geschichtsmetaphysik des Rolandsliedes und ihre Vorgeschichte,” Zeitschrift für romanische Philologie, 55 (1935), 1-87; and Michael Wendt, Der Oxforder Roland (Munich, 1970), pp. 170ff.

9. See George Fenwick Jones, The Ethos of the Song of Roland (Baltimore, 1963), p. 121.

10. See Herman Gräf, “Der Parallelismus im Rolandslied,” dissertation Julius-Maximilians-Universität, Würzburg (Wertheim a. M., 1931); Wendt, Der Oxforder Roland, pp. 179-211.

11. See Paul Rousset, “La croyance en la justice immanente à l'époque féodale,” Le moyen-âge, 54 (1948), 225ff.

12. See, for example, the interesting debate among Larry S. Crist, Wolfgang G. van Emden, and William W. Kibler in the 1974 and 1975 volumes of Olifant.

13. Joachim Bumke, Wolframs Willehalm (Heidelberg, 1959), pp. 57ff. Compare Hans Robert Jauss, “Chanson de geste et roman courtois au XIIe siècle,” in Chanson de Geste und höfischer Roman, Heidelberger Kolloquium, January 30, 1961 (Heidelberg, 1963), pp. 61-77; translated by the present writer.

14. Menéndez Pidal, La Chanson de Roland, p. 504.

15. See Karl-Joseph Steinmeyer, Untersuchung zur allegorischen Bedeutung der Träume im altfranzösischen Rolandslied (Munich, 1963); W. G. van Emden, “Another Look at Charlemagne's Dreams in the Chanson de Roland,” French Studies 28 (1974): 257-71.

16. Compare Bruce A. Rosenberg, Custer and the Epic of Defeat (University Park, Pa., and London, 1974).

17. See Menéndez Pidal, La Chanson de Roland, pp. 165ff, especially 167.

18. See lines 755ff, 790ff, 806; Menéndez Pidal, La Chanson de Roland, p. 432.

19. See Wendt, Der Oxforder Roland, pp. 281ff.

20. See Waltz, Rolandsliedslied, pp. 110ff; L. C. Macinney, “The People and Public Opinion in the Eleventh-Century Peace Movement,” Speculum, 5 (1930), 181-206. The peace movement led by the Church was an attempt to get the nobles to stop fighting among themselves.

21. Uitti, (see Bibliography, below), p. 99.

22. See F. L. Ganshof, Feudalism, trans. Philip Grierson, 2d English ed. (New York: Harper Torchbooks, 1961).

23. See Frederick Goldin, “Die Rolle Ganelons und das Motiv der Worte,” Zeitschrift für romanische Philologie, forthcoming.

24. See Bender, König und Vasall, p. 33f.

25. See Köhler, “Jugement des Barons.”

26. Saint Augustine, Tractatus CXXIV in Joannis Evangelium, Tractatus XXVII, P. L. 1619f.

27. See the following works: Milman Parry, The Making of Homeric Verse: The Collected Papers of Milman Parry, ed. Adam Parry (Oxford, 1971).

    Jean Rychner, La Chanson de geste: Essai sur l'art épique des jongleurs (Geneva and Lille, 1955).

    Albert B. Lord, The Singer of Tales (Cambridge, Mass., 1960); reprinted in paperback by Atheneum, New York, 1973.

    Stephen G. Nichols, Jr., Formulaic Diction and Thematic Composition in the Chanson de Roland (Chapel Hill, N.C., 1961).

    Edward A. Heinemann, “Composition stylisée et technique littéraire dans la Chanson de Roland,” Romania, 94 (1973), 1-28. See also the same author's review of Duggan's book in Olifant, 1 (October 1973), 23-31.

    John Miletich, “The Quest for the ‘Formula’: A Comparative Reappraisal,” Modern Philology, 74 (November 1976), 111-23. See also the same author's article, “Narrative Style in Spanish and Slavic Traditional Narrative Poetry: Implications for the Study of the Romance Epic,” in Olifant, 2 (December 1974), 109-128.

28. See Lord, Singer of Tales, p. 95.

29. See Italo Siciliano, Les Origines des chansons de geste, tr. P. Antonetti (Paris, 1951). The monumental statement of the individualist position is the four-volume work by Joseph Bédier, Les Légendes épiques, published in Paris from 1926 to 1929. This remains a vital and indispensable work, though the argument has lost credit. One should also consult his Commentaires to his translation of the poem, published in Paris in 1927.

30. A poem “that lives in variants:” this phrase is taken from the title of Chapter 2, “Une poésie qui vit de variantes,” of the work by Menéndez Pidal, La Chanson de Roland.

31. Ibid., p. 63f.

32. The only other assonanced version is contained in a manuscript in the library of San Marco in Venice, designated V4. Written in a mixture of French and Italian, the manuscript dates from the fourteenth century, though the composition of the poem is much earlier. In its first 3845 lines this version is close to the first 3681 lines of the Oxford version; it then goes on to relate other episodes to the end, line 6011.

    There are, in addition, seven rimed versions (of which two are fragmentary) dating from the end of the twelfth and the thirteenth century. These versions recast the old song into rimed couplets and expand the material, adding much that betrays the influence of the courtly romance.

    There were also translations and recensions of various versions of the story into German, Norse, Welsh, English, Dutch, Spanish, and other languages. Of these, the greatest is the Ruolandes liet of Pfaffe Konrad, composed either in the 1130s or around 1170 (each date has its defenders). For an account of all of the versions of the story of Roland, see de Riquer, Chansons (see Bibliography, below), pp. 52-59.

33. See Rudy S. Spraycar, “La Chanson de Roland: An Oral Poem?” Olifant, 4 (1976), 63-74. For a balanced and thorough examination of this question, see Maurice Delbouille, “Les chansons de geste et le livre,” in La technique littéraire des chansons de geste, Actes du Colloque de Liége, Bibliothèque de la Faculté de Philosophie et Lettres de l'Université de Liège, Fascicule 150 (Paris, 1959), pp. 295-407.

34. See Lord, Singer of Tales, pp. 25ff.

35. On oral poetry as a system for preserving and retrieving the wisdom of the past, see Walter J. Ong, S.J., “World as View and World as Event,” American Anthropologist, 71, 4 (August 1969), 634-47, especially pp. 638-641 and the bibliography at the end of the article.

36. See Paul Zumthor, Langue et techniques poétiques à l'époque romane (XIe-XIIIe siècles) (Paris, 1963), pp. 27-69.

37. See Robert A. Hall, Jr., “Linguistic Strata in the Chanson de Roland,” Romance Philology, 13 (1959-60), 156-61.

38. In the laisses similaires (for example, laisses 40-42, 83-85), the action stops completely; in the laisses parallèles (for example, laisses 71-78, 93-95, 96-103, 218-225, 232-34), the action is advanced. See Rychner, La chanson de geste, pp. 83-107.

39. Exceptionally beautiful examples of this process of chronicling and laudation are in lines 2083-93, and 1678-85. See Frederick Goldin, “Le noyau temporel de la laisse dans la Chanson de Roland,” forthcoming in the Actes of the Eighth Congress of the Société Internationale Rencesvals.

Bibliography

For an introduction to the poem and its literary and historical background, the following are recommended:

Jules Horrent. La Chanson de Roland dans les littératures française et espagnole au moyen-âge. Paris, 1951.

Pierre Le Gentil. La Chanson de Roland. 2d ed. Paris, 1967.

———. The Chanson de Roland, tr. Frances F. Beer. Cambridge, Mass., 1969.

Martín de Riquer. Los Cantares de gesta franceses. Madrid, 1952.

———. Les Chansons de geste françaises. 2d ed. tr. Irénée Cluzel. Paris, 1968.

D. Karl Uitti. Story, Myth, and Celebration in Old French Narrative Poetry, 1050-1200. Princeton, N.J., 1973.

Eugene Vance. Reading The Song of Roland. Englewood Cliffs, N.J., 1970.

Studies of the poem are listed in the standard bibliographies (such as those by Robert Bossuat, Urban T. Holmes, John H. Fisher). A good critical survey of earlier scholarship can be found in the following articles by Albert Junker:

“Stand der Forschung zum Rolandslied,” Germanisch-Romanische Monatsschrift 37 (1956): 97-144.

“Von der Schönheit des Rolandsliedes (O) im Spiegel neuester Forschung,” in Medium Aevum Romanicum, Festschrift für Hans Rheinfelder, ed. H. Bihler and A. Noyer-Weidner. Munich, 1963.

Critical bibliographies of the most recent work on the poem can be found in The Year's Work in Modern Language Studies, and especially in the Bulletin bibliographique de la Société Rencesvals. A lively and useful review of recent work and many important contributions are published by the American-Canadian branch of the Société Rencesvals in its quarterly, Olifant.

On the story of Roland in art, see Rita Lejeune and Jacques Stiennon, La Légende de Roland dans l'art du Moyen Age, 2 vols. (Brussels, 1966); translated by Christine Trollope, The Legend of Roland in the Middle Ages (London and New York, 1971).