¹ La Chanson de Roland (Boston, 1924), p. xxxviii.

² “Four Chansons de Geste: A Study in Old French Epic Versification”, MLR, viii (1913), 352-367.

³ La Chanson de Roland commentée par J. B. (Paris, 1927), pp. 117-118.

⁴ “Reconstruction of the Original Chanson de Roland”, Trans. Connecticut Acad, of Arts and Sciences, xv (1909), 111-137; Romanic Review, iv (1913), 112-117.

⁵ French Precursors of the Chanson de Roland (New York, 1948), pp. 36-41.

⁶ La Chanson de Roland (Paris, 1937).

⁷ To the present writer the “Baligant episode” is composed of that plot element where the units-of-three are not a characteristic structural device. The “Marsile episode” is that plot element where the units-of-three are an integral and basic technique of the structure. The lines of demarcation between the two parts are difficult to determine for the author of the Baligant episode was almost as good a poet as the author of the Marsile episode and evidently did his utmost to interpolate the addition into the poem so that the Chanson form an apparently integrated whole. The Marsile episode is all of the first part of the poem up to Charles' arrival on the field of battle, his pursuit of the pagans and at least one of the dreams (laisses cxxxxv, cxxxxvi), the latter dream probably being the one which was in the original poem. That much of the plot material forms a story unit and is characterized by the use of the triple rhythm. The Baligant episode has begun certainly by laisse cxxxxvii, for there the three pagan Gods are mentioned as a Trinity for the first time. As explained later in this study this trio of Gods, as a Trinity, is avoided scrupulously in the Marsile episode. The first interpolation ends with line 2844. Line 2845, “Al matin, quant primes pert li albe”, originally must have followed line 2569, “Carles se dort tresqu'al demain, al cler jur”, though the laisse in which this line appears was probably rewritten, for this line itself is defective. In laisses cxxxxv-vi the angel Gabriel is guarding Charles and in laisse cciii he is still watching over him.

Lines 2845-2974 must have been in the original poem for we again find the device of three. Charles recognizes the blows of Roland on three stones, he has three men prepared for burial. In his five-fold lamentation over Roland Charles develops three themes: (1) he prays for God's mercy on Roland; (2) he laments the loss of such a loyal friend and warrior; (3) he foresees the decline of his own honor and power. The latter two themes are in all five of the laisses, but significantly enough, the prayer for God's mercy is found in only three.

The second part of the Baligant episode begins at line 2974 and continues to approximately line 3675. In lines 3689-91 the poet mentions the burial of the three men. It is possible that laisses ccxii, ccxiii, and the original of laisse ccxxvii were a unit of three strophes relating the burial of the troops and of the three heroes, and Charles' return to France.

The trial and execution of Ganelon, where we find again the three-fold rhythm, are the conclusion of the Marsile episode. References here to Bramimonde can be explained as additions by the author of the Baligant interpolation. The final laisse, with Charles' dream, is the concluding element of the three-fold dream device.

An effort to find evidence of the units-of-three technique results in the following outline of lines 2570-2844 as typical of the Baligant episode: clxxxvii, Marsile goes to Saragossa, he insults the pagan Trinity; clxxxviii, lament of Bramimonde, she speaks of the émir; clxxxix, a flashback to the message sent to Baligant seven years earlier; cxc, Baligant sails for Spain; cxci, Baligant sails up the Sebre; cxcil, Baligant lands. He threatens Charles; cxcni, he instructs messengers to Marsile; cxciv, the messengers set out and arrive at the palace; cxcv, the messengers enter the palace and greet Marsile. Bramimonde tells the bad news; cxcvi, they deliver the message. Bramimonde answers that Charles is nearby; cxcvii, Marsile renders Spain to Baligant and sends him the keys of Saragossa; cxcviii, Marsile tells where Charles is, the messengers leave; cxcix, Clarien reports to Baligant about Marsile's defeat; cc, Clarien also reports the disaster of the French. Baligant gives orders to his troops; cci, the pagans go to Saragossa. Bramimonde greets Baligant; ccn, conversation between Baligant and Marsile, Baligant accepts the glove, the pagans mount and ride away. Any effort to group these strophes into units-of-three fails. A survey of the remainder of the Baligant episode produces a similar result except for the two units found in lines 3214-64 and 3369-3404.

⁸ In volume iii of his Les Légendes Epiques (pp. 393 ff.) Bédier gives a succinct résumé of the many arguments for multiple authorship and an answer to them all. He states (pp. 398-399) that the only legitimate basis for such arguments is “celui qui, s'attachant à la critique interne du texte, essaye d'y découvrir des disparates, soit dans la marche du récit, soit dans la peinture des caractères, et s'efforce de mettre le ‘dernier rédacteur’ en contradiction avec lui-même.” He refutes all arguments of this nature which have been advanced. If the argument and evidence of the present study be valid, then such a “disparate” has been found in the use and absence of the triple rhythm, and it leads inevitably to the conclusion of dual authorship.

Knudson, in Romania, ixiii, 48-92, argues that the unity of the poem is the author's “dessein de servir à la gloire de Charlemagne”, and that the concern for Charlemagne's glory was the “idée maîtresse” of the poet. If such was the purpose of the “single author” then he failed, for the poem remains a work in which Roland is the central and predominant character. The poet was too great a genius to have failed to make Charlemagne the central figure of the whole work, if such had been his intention.

In an unpublished thesis at Ohio State University Miss Norma Bondy (“Thematic Repetitions in the Chanson de Roland”, 1945) studied the repetitive themes which occur as leitmotivs throughout the work. She concluded that the repetitions are a conscious stylistic device, that they constitute structural unity, and that the poem is therefore the work of one author. Her conclusion of single authorship is based upon thirteen cases of verses of the first part which are repeated in the “interpolated divisions” of the poem.

(There are thirteen lines repeated, but only nine themes for some of her repetitions cover more than one line.) These repetitions can just as well be explained as a deliberate effort on the part of the second poet to imitate the first poet. Lifting lines from the first part would be a perfectly natural procedure.

Further evidence that the poet of the second part was trying to achieve an appearance of unity through imitation can be seen in a study of the dream sequences. Of the five dreams which Charles has, four are specific and their symbolism is obvious. Looking at them a posteriori their meaning is evident. However, the dream in laisse clxxxv seems to be spurious. It lacks the precision and obvious symbolism of the other four dreams and it is difficult to decide to what it refers. It must represent an unsuccessful attempt to imitate the original poet's technique of forecasting the action through dreams.

⁹ A. H. Schutz (MLN, lxii, 456-461) has discussed the congé of this unit and concludes that it is a “missio” and the leave-taking is comparable, though in the nature of the case not identical, with a monastic Itinerarium. The use and meaning of the number symbolism here seem to strengthen his suggestion that this is a dismissal ceremony with the appropriate paternal blessing, and that as the lay ceremony approaches the religious rite, the situation becomes more dramatic.

¹⁰ There is a slight suggestion that five means disaster for the Christians. Ganelon is the fifth and final name mentioned as emissary for Charles. Unit 10 is in five laisses. The fifth pagan assault is the worst (1. 1687). Charles' lament over Roland is in five laisses (11. 2881-2944).

¹¹ L'Art religieux du treizième siècle en France (Paris, 1927), pp. 5-26.