Medieval Spanish epic cycles: observations on their formation and development

Alan Deyermond

The epic traditions of medieval Spain are, it is generally agreed, closer in some ways to historical reality than are the epics of France or most other countries. Whether one considers the extant verse texts, the epic legends incorporated in the prose of Latin and vernacular chronicles, or the ballads composed on epic subjects, it is often possible to find characters -even minor characters- whose existence is independently attested in legal documents, echoes of events long past, references to institutions, and assumptions about the political or military situation which are explicable only in terms of a period long before that of the extant text. The frequent occurrence of historically accurate material coincides with an unusual restraint in the use of the supernatural and the marvellous, and with an even more unusual care to avoid the grosser forms of factual exaggeration (the numbers of enemies killed by Spanish heroes are exaggerated, but remain within the bounds of plausibility). Scholars disagree about the causes and the implications of these features, but their existence has long been accepted. It was the object of sporadic comment in the century which followed Tomás Antonio Sánchez's publication of the Cantar de Mio Cid (1779), was developed as a coherent theory by Manuel Milá y Fontanals in De la poesía heroico-popular castellana (Barcelona: Álvaro Verdaguer, 1874), and was demonstrated much more extensively in a series of studies by Ramón Menéndez Pidal¹.

The difference between the Spanish and other epic traditions may well be less than is now supposed: scholars, like other people, tend to find what they are looking for, and generations of Spanish investigators have been trained to look for historical elements in the epic, whereas similar elements in the epics of other countries may more often have passed unnoticed. Yet, even if one makes a generous allowance for the effect of this psychological factor, it is almost certainly true that the traditional Spanish doctrine is correct, and that Spanish epic really does contain a surprising amount of historical material.

The recognition of this fact has brought obvious benefits to epic studies, but these are inevitably balanced by disadvantages. First, there has been a tendency to assume that, because much is historical, all must be -or, if the extant texts do not support such an assumption, that their lost originals were wholly accurate historically. There is, of course, no evidence whatever for this belief, and there is a good deal of evidence against it. Secondly, the search for historical material has caused scholars to neglect aspects which are both interesting and important. This attitude is very frankly stated by Menéndez Pidal, who had for long abandoned his intention of studying La condesa traidora because «resultaba asunto estéril para el estudio de la epopeya como poesía de tema histórico y de vida tradicional»². Thus the literary qualities which make the epics interesting have been largely neglected. A brilliant essay by Dámaso Alonso, published in 1941, encouraged literary criticism of the Cantar de Mio Cid³, and it is now, fortunately, taken for granted that this is a useful and natural way in which to study the Cantar, but only recently have other epics benefited from literary rather than historical research. Two of the leading hispanists of this century -William J. Entwistle in 1933, and María Rosa Lida de Malkiel in 1949⁴ -pointed out the absurdity of this neglect, but traditional assumptions were so strong that their pleas were little heeded.

Clear signs of a healthier balance of interests may now be seen. In addition to much criticism of high quality on the Cantar de Mio Cid (the outstanding achievement so far is Edmund de Chasca's El arte juglaresco en el CMC) and to a useful body of criticism on the Poema de Fernán González (notably in articles by J. P. Keller, Joaquín Gimeno Casalduero, and Juan Bautista Avalle-Arce), two recent articles have shown that it is possible to study the lost epics in some detail from a literary point of view: Charles F. Fraker has worked on the Cantar de Sancho II, and J. G. Cummins on the Siete infantes de Lara⁵. It seems to me that studies of this kind may contribute usefully to the discussion of epic cycles, and that literary analysis may equally benefit from work done on the formation of cycles and on the relationship of one epic poem to another.

Scholars disagree vigorously and frequently on most aspects of the Spanish epic: the number and nature of lost poems, the dating and authorship of both extant and lost poems, the part played by oral composition, the principles underlying epic versification, and so on. On some points, however, there is, if not unanimity, at least a consensus. Most scholars accept that the plots, and occasionally some lines, of lost epics are preserved by chronicles and ballads from the thirteenth to the sixteenth century; that there was a group of poems on events in the reigns of the counts of Castile, and another group on the deeds of the Cid; and that, whether or not there were poems on events before the mid-tenth century, the Siete infantes de Lara is the first epic whose date of composition can be fixed by firm evidence (ca 1000 seems an acceptable latest date)⁶.

In this article, I shall consider the nature of the epic cycle on the counts of Castile and events in their reigns, and shall attempt to identify the distinctive features of that cycle. I shall then discuss borrowings and reminiscences within the cycle (e. g. the debt of the Romanz del infant García to the Siete infantes), as well as the cycle's borrowings from outside, and its influence on later epics. Finally, I shall consider whether the Cantar de Sancho II may not perhaps have begun as a member of the Counts of Castile cycle. It will, I think, become clear that if one considers the poems as literary creations (literary in the broadest sense -I am not concerned at this point with the debates on clerical versus secular authorship, and on written or oral composition), the relations between them differ from those that one might expect from a study of their historical content.

The properly attested poems of the Counts cycle are those referred to above. On the first count, Fernán González, we possess most of the Poema de Fernán González, a reworking in the learned cuaderna vía metre of the lost Cantar de Fernán González, of which traces may be found in fourteenth-century chronicles and ballads. The second count, Garci Fernández, is the central though not the dominant character of La condesa traidora (plot fully given by the chroniclers), and the Siete infantes (many lines preserved, and plot fully given, by the chroniclers; several ballads) is set in his reign. Sancho, the third count, is not the protagonist of any poem which can be traced, but he plays a prominent role in the conclusion of La condesa traidora, and in the motivation of the Romanz del infant García. The latter poem (plot fully given by the chroniclers, who in one case mention the poem's title and compare its content with what the Hispano-Latin historians say) tells of the murder of the fourth and last count of the native Castilian line. There may well have been other poems in the cycle, but if so, they have not survived either in verse texts or in chronicle prosifications, with the possible exception of the Abad don Juan de Montemayor, which is first mentioned so late that its origin and even its genre are doubtful⁷. It is probable that these poems existed in more than one version, though some scholars believe that the differences between one chronicle account and another are due not to the use of different poetic sources but to changes introduced for either artistic or historical reasons by the chroniclers themselves. In either case it is inadvisable to combine lines or plot elements from several chronicles, as has sometimes been done, since the result would be a text which, however interesting, was a modern construct unrepresentative of the versions which circulated in the Middle Ages. Each chronicle version should therefore be treated independently, and each, even if some of its elements are the work not of a poet but of a chronicler, can safely be regarded as an artistically coherent realization of an epic tradition.

The common feature of these poems which has hitherto been most often stressed is, as we have seen, their historical content, but although this is real enough, it varies so much in intensity that it can scarcely be regarded as a major characteristic of the cycle. The Romanz del infant García seems to have the highest degree of historical accuracy, though even here history is subordinated to fiction in the construction of the plot. In the Poema de Fernán González, the historical content is smaller but still substantial, and in La condesa traidora and the Siete infantes there is very little historical reality, while the plot outlines belong to the familiar stock of traditional fiction.

A more striking and more uniform feature is the strong connection between these poems (at least, in the forms in which they have come down to us) and the tomb cults of medieval Spanish monasteries and churches. The position was accurately set out by P. E. Russell almost twenty years ago⁸, and as far as I am aware his statements have not been challenged (it would, in fact, be impossible to challenge them, since they simply juxtapose two sets of data which may be easily checked). The version of La condesa traidora in the Crónica Najerense has as its dénouement the rescue of Garci Fernández's body from Moorish territory by his heir, Sancho, and its burial at San Pedro de Cardeña. In the fuller version found in the Estoria de España the climax is a fictitious account of Sancho's foundation of San Salvador de Oña after his treacherous mother's death. The Romanz del infant García, as represented in EE, tells of the murdered count's burial at Oña (where, as with the rival tomb in León favoured by the Hispano-Latin thirteenth-century chroniclers, an epitaph preserves the story of the murder). Two monasteries also competed in displaying tombs of the Siete infantes de Lara: San Pedro de Arlanza and San Millán de la Cogolla each had seven tombs allegedly containing the headless corpses, and what were claimed as the infantes' skulls were proudly displayed at the parish church of Salas de los Infantes. To all of this we may add the well-known fact that Fernán González founded the monastery at Arlanza and was buried there; the Poema, which records his donations to the monastery and his dependence on the monks for advice and comfort, is generally agreed to be a mid-thirteenth-century product of Arlanza⁹. It does not follow that the epics are all ecclesiastical compositions, inspired by the tomb cults; as Russell notes (p. 58), we have no means of arriving at a reliable conclusion. Moreover, it seems likely, for several reasons, that the epic of the Siete infantes inspired the rival cults, and not vice versa. Nevertheless, the connection with tomb cults deserves much more attention as a common feature of this cycle than it has so far received.

A third feature -and one which lies closer to the centre of the poems' nature- is the part played by vengeance. This is least noticeable in the Poema de Fernán González, and Eugene Dorfman has difficulty in bringing the Poema within what he believes to be the dominant pattern of French and Spanish epic (family quarrel-insult-treachery-punishment)¹⁰. In the other three epic traditions of the Counts cycle, there is greater emphasis on vengeance, and in at least two cases on counter-vengeance. In García (EE and Cr1344 versions), the Vela brothers avenge their humiliation at the hands of Count Sancho by the treacherous murder of his son, which is in turn avenged by García's fiancée. In the Cr1344 version of Siete infantes, Álvar Sánchez, Doña Lambra's cousin, insults Gonzalo González and his brothers; the insult leads Gonzalo to strike the offender who, by ill fortune, dies from the blow; Lambra plans one revenge which miscarries, and then persuades her husband Ruy Velázquez to arrange the killing of the infantes and their father; and, in the fullness of time, the dead infantes are avenged by their half-brother Mudarra, begotten by Gonzalo Gústioz in prison after his narrow escape from death. La condesa traidora has, in the very similar versions given by EE and Cr1344, two acts of vengeance, when Garci Fernández kills his faithless wife and her lover, and when Sancho compels his mother to drink the poison she had prepared for him. There is not, however, a chain of revenge -unless, of course, one regards Garci Fernandez's two wives (both French, both the daughters of counts who passed through Castile on pilgrimage, both faithless, both killed) as in some sense identified with each other; and such an identification would, despite its attractions, raise a good many problems.

Vengeance is, then, a major theme in three out of the four epics, and it plays a considerable part in the fourth (PFG). We must remember, however, that it is important in the epic of widely separated lands and periods, so we should not attach excessive weight to its occurrence in the Counts cycle. The chain of vengeance in Siete infantes and in García is a more significant agreement.

Of even greater significance is the association of vengeance with women characters who play an active, and sometimes a dominant, part. Such characters are to be found from time to time in the epics of other countries, perhaps the best known being Brunhild and Kriemhild in the Nibelungenlied, but they are more frequent in other areas of literature, notably the folktale and the romance. In the epic, it is more usual to find that women play hardly any part (as in the Chanson de Roland), or that their role, even though essential to the motivation of the plot, is largely passive (as with Helen in the Iliad, or Ximena and her daughters in the Cantar de Mio Cid). The only Spanish epics outside the Counts cycle in which a woman takes an active part are the Mocedades de Rodrigo (Ximena determines to avenge her father's death, but ends by asking for Rodrigo as her husband, after which her role becomes passive) and the Cantar de Sancho II, which, as we shall see, may have been conceived as a late member of the Counts cycle. Yet what is exceptional elsewhere is the norm in that cycle.

In the Poema de Fernán González, the hero's capture and imprisonment in Navarre are caused by the treachery of Teresa, Queen of León. He is released by the Navarrese princess Sancha, who rapidly proves herself a woman of courage and resourcefulness (she stabs an archpriest who tries to rape her during the escape), and after she and Fernán González are married she shows great daring in bringing about his escape from a second imprisonment.

The title which scholars have given to La condesa traidora indicates the dominant role of Sancha, Garci Fernández's second wife, in the latter half of the epic: inflamed by an illicit love for the Moorish king, she conspires with him to cause her husband's death, and later tries to poison her son because he too is an obstacle to her desires. Garci Fernández's first wife, Argentina, resembles Sancha in many ways, as we have seen, but although she is important in the first half of the epic she responds to the initiatives of others rather than dominating the action. She is no match for her de facto stepdaughter Sancha, who, illtreated by her, takes revenge by arranging the murder of Argentina and of her own father by the enraged Garci Fernández¹¹. In the Siete infantes the tragic chain of events, from the insult at the wedding to the killing of the seven brothers and the attempt to have their father killed by the Moors in Córdoba is, as we have seen, the work of Doña Lambra, either directly or as the result of her incitement of her husband. The version of the story given by EE deals in a relatively perfunctory way with the vengeance exacted by Mudarra (so perfunctory, indeed, that one is inclined to suspect compression by the Alfonsine chroniclers), but in Cr1344 we find a second part of the story worthy of the first in its scope and tragic power. Not only does Mudarra play a much larger part, as Cummins shows¹², but Sancha, wife of Gonzalo Gústioz, takes over the direction of events. She joyfully welcomes her embarrassed husband's bastard when the youth arrives in Castile, since she sees in him the only hope of revenge; thanks to her he is accepted (she adopts him as her own son), and is knighted by Garci Fernández, thus becoming entitled to challenge the traitor Ruy Velázquez; and she is in at the kill, deciding Ruy Velázquez's fate and even trying to drink his blood (Cr1344, chap. 379, p. 169). Sancha thus balances Lambra, directing the revenge just as Lambra had directed the treachery, and she becomes as savage as Lambra -the loving mother and wife has become an avenging fury, whose desire for blood shocks even the warrior hero Mudarra:

E entõ pos os geolhos em terra pera lhe bever do sangue. Mas dom Mudarra Gonçálvez a tomou pello braço e alevantouha da terra, dizendo: -Nom queira Deus, madre senhora, que tal cousa passe, que sãgue de homen assý treedor entre corpo tam leal e bõo como o vosso he!

(p. 169)

Erich von Richthofen, discussing the analogies between this episode and a vengeance scene from the Norse Ragnarssdrápa, concludes that «De toda la épica europea son éstas dos las escenas más sangrientas»¹³. Sancha thus develops in much the same way as Kriemhild who, in the Nibelungenlied, begins as a gentle maiden but is turned by long brooding on the unavenged murder of her husband into a monster of destruction¹⁴. Lambra and Sancha, then, dominate the epic of the Siete infantes, but one other woman plays an important part. In the EE version, a Moorish noblewoman is sent by Almanzor to look after Gonzalo Gústioz in prison, they become lovers, and the Mooress conceives Mudarra (Primera crónica general, chap. 738, p. 435a): this happens before Gonzalo learns of the death of his sons. In Cr1344, on the other hand, the Mooress (who in this version is Almanzor's sister) comforts Gonzalo after he has wept over the heads of his sons, and she provokes him into making love to her:

E eu véjovos os cabellos brancos e o rostro muy fresco e muy fremoso; e per ventuira podees ainda fazer filhos que vingarõ os outros.

(chap. 376, p. 149)

It is clear that, despite her modest protests when Gonzalo takes hold of her, the Mooress has taken the initiative in the conception of an avenger. She is thus at the central point of the epic's structure, and in Cr1344 seems conscious of her role, and as decisive, though not as fierce, as Lambra or Sancha.

A development similar to that of Sancha in the Siete infantes may be seen in her namesake in the Romanz del infant García. The Leonese princess, a girl happily in love, witnesses the murder of her fiancé, and is herself brutally treated by one of the murderers. All but one are quickly caught and put to death («Los reys cercáronlos estonces a los otros condes [the Vela brothers], et quemçaronlos ý luego, faziéndoles antes muy grandes penas como a traydores que mataran a su sennor», PCG, chap. 789, p. 472a). The survivor is punished only thanks to the persistence of Sancha who, betrothed to Prince Fernando of Navarre (the future Fernando I of Castile), swears that she will never consummate the marriage until the remaining murderer is hunted down. He is caught, and Sancha, asked like the Sancha of the Siete infantes to decide on a suitable punishment:

tomó un cuchiello en su mano ella misma, et tajóle luego las manos con que él firiera all inffant et a ella misma, desí tajól' los pies con que andidiera en aquel fecho, después sacóle la lengua con que fablara la trayción...

(PCG, p. 472b15)

It will be seen that the punishment inflicted by Sancha with her own hands is more savage than that suffered by the other murderers at the hands of the King of Navarre and his sons, and that the description is much more detailed. In every epic of the Counts cycle which has come down to us in verse or prose, part at least of the action is strongly influenced by a woman: the second half of the epic in Fernán González, Condesa traidora, and García; the first half of Siete infantes as incorporated in EE; and the entire plot in the Cr1344 version of the same epic. In two of the epics (Fernán González and the Cr1344 version of Siete infantes) part or all of the action turns on the struggle between two women, one for the hero(es) and the other against. The women generally have dominant personalities, and several prove capable of great savagery.

The bloodthirstiness of the two Sanchas, in Siete infantes and García (which arises, as in the Nibelungenlied, not only from the outrages they have suffered but also from the necessary postponement of revenge), seems to have shocked the compilers of Cr1344 or the poets on whom they drew. Mudarra, as we have seen, prevents his mother from drinking Ruy Velázquez's blood, but a comparison of this scene with Sancha's dream of vengeance («E a m parecia que per elle [Ruy Velázquez's body] corryã rios de sangue; e eu punha os gyolhos em terra e bevya do sangue delle», Cr1344, chap. 378, pp. 156-57) suggests strongly that in an earlier version she really did drink the traitor's blood. In the case of García, supposition is unnecessary, for we are able to make a direct comparison between the EE and Cr1344 versions of Princess Sancha's vengeance. In the former, she herself carries out the mutilation of the murderer Fernán Laínez, whereas in the latter: «Entõ a iffante dona Sancha tomou huũ cuytello em sua mãao e cõ elle mãdou que lhe cortass as maãos...» (chap. 428, p. 251). This toning down of a woman's direct and personal savagery parallels the removal in Cr1344 of the erotic note found in the EE account of the meeting between García and his fiancée. Menéndez Pidal argues that the bloodthirstiness of the two Sanchas is a late development:

sólo en ulteriores desarrollos de la leyenda [of the Siete infantes] en el siglo XIV se introducen los rasgos de ferocidad femenina que acusan una decadencia en el gusto épico. Generalmente se cree que la mayor barbarie en las costumbres de un relato tradicional es signo de su mayor antigüedad, pero el estudio de la evolución de los relatos épicos medievales contradice esta idea vulgar16.

The evidence I have cited is admittedly not conclusive, but it does cast a good deal of doubt on Menéndez Pidal's view. It seems more likely that the savagery of some women characters is present in the Counts cycle from an early stage. However, it is not necessary to accept Lucy A. Sponsler's suggestion that assertive and even ferocious women characters may reflect «the rigor of life in early medieval Spain»¹⁷. These are, like many other characters and motifs in the Spanish epic, a feature of universal folk narrative¹⁸, and epic poets make their individual selections from the common stock. The noteworthy point in the context of the present study is the extent of the agreement between the four poems of the Counts cycle, and especially between García and the Cr1344 version of Siete infantes.

A careful comparison of any two epics, even ones widely separated in time and language, will reveal points of similarity: for example, recent critics of Gilgamesh and of the Cantar de Mio Cid have argued, rightly, that the education of a king in proper conduct is an important theme of their poems¹⁹. Sometimes such resemblances may be due, as we have just seen, to common membership of an epic cycle. In other cases, they may represent independent use of underlying patterns. Finally, they may be the result of direct borrowing by one epic poet from another.

The work of Parry and Lord has accustomed us to the idea that oral epic poets have a common stock of formulas, and that detailed verbal similarities need not imply direct borrowing from one particular poem²⁰. Equally, similarities in larger narrative units -motifs, or even themes²¹- may arise from use of a common stock, though one that is sometimes less easy to define than a formulaic system. One example of such underlying patterns is the influence, whether conscious or unconscious, of liturgical motifs: John K. Walsh has shown how several Spanish epics make use of martyrological material from the liturgy²². A second is the use of folk motifs, which is much more frequent in Spanish epic -and no doubt in that of other countries- than might be supposed²³. Other possible patterns lie deeper and are thus harder to study, but Adrian G. Montoro has shown the relevance to Spanish epic of a concept from Georges Dumézil's variant of structural anthropology: the three functional levels, in primitive Indo-European peoples, of royal-divine, warlike, and economic-sexual-agricultural²⁴. This approach, at first sight modishly eccentric, proves serious and disconcertingly valid. Basic patterns of myth have been shown by Peter N. Dunn to affect the themes and structure of the Cantar de Mio Cid²⁵, and a similar examination of other Spanish epics might well yield useful results. Finally, and perhaps even more disturbingly, Lord raises the possibility that oral epics (and therefore, one might add, epics composed in writing but dependent on an originally oral tradition) are «traditional units with their parts belonging together by a kind of mythic necessity or by thematic attraction» (p. 206). He goes on to suggest that the startling parallels in plot which he discovers between Beowulf and the Chanson de Roland are to be explained by such an attraction of motifs. Other scholars have pointed to other parallels of this sort: for example, between the story of Achilles in the Iliad and that of Sigurd in the Norse Volsungasaga²⁶, and between the murder of Siegfried in the Nibelungenlied and that of Sancho in the Cantar de Sancho II²⁷. Margaret Chaplin and I have argued elsewhere that the killing of Sancho with his own weapon is an example of a widely diffused motif²⁸, but other patterns may also be at work. Why does Sancho go off alone with the Zamoran knight Bellido Adolfo, against whom he has already been warned, and then hand him his spear? This is grossly imprudent behaviour on the part of a king who is besieging Zamora, and poet or chroniclers seem to have felt the need to justify it. EE, with which Cr1344 agrees at this point, explains it as a reconnaissance carried out on Bellido Adolfo's terms (and thus potentially vital to the success of the siege), the king's dismounting as due to a desire to enjoy a walk along the river bank, and his final imprudence in these words:

Et trayé en la mano un venablo pequenno dorado commo lo avién estonces por costumbre los reys, et diol'a. Vellid Adolfo que gele toviesse, et el rey apartósse luego a fazer aquello que la natura pide et que ell omne non lo puede escusar29.

This anxiety to rationalize a puzzling action suggests -as in other versions of traditional stories- that in an earlier version suprarational forces may have been seen at work. The Crónica Najerense, as usual, gives a much shorter account:

regem extra castra iusta muros quasi ad explorandum urbis introitum deduxisset et rex de equo descendens, et nature sederet necessária, ipse super alterum equum insidens, emisso eum uenabulo, interfecit.

(Book III, § 42, p. 114)

It seems that the Najerense account is an abbreviated version of that used by EE, though the lines of the lost Carmen de morte Sanctii regis that have been reconstructed from the Najerense prose³⁰ suggest that the Latin poet was following a different account, and one much closer to what actually happened -for we must remember that Sancho's assassination was a historical reality. The evidence is tantalizingly incomplete, but it seems to indicate that by the middle of the twelfth century a Cantar de Sancho II had recast that reality in terms of a traditional pattern³¹. That pattern may well have been one of ritual sacrifice, perhaps even with the consent of the victim, since Sancho's imprudence is otherwise hard to understand³². It is not possible to say whether the rationalization that is evident in the Najerense account was an early reaction by a chronicler, followed and expanded by the Alfonsine chroniclers, or whether it was already present in the Cantar, in which case it may have formed an integral part of a half-conscious adoption of the sacrificial pattern. A final question that must be asked about this episode -as in all cases where «mythic necessity» is invoked- is whether this may not be a straightforward literary borrowing, in this case from the Nibelungenlied. Erich von Richthofen believes that it is, but if the episode had in the twelfth-century Cantar much the same shape as in the EE version, then the Nibelungenlied, composed about 1200, cannot be the source³³. It is possible that a lost German ancestor of the Nibelungenlied was used by the poet of Sancho II, but it seems to me more likely that the resemblance of the historical assassination to part of a mythic pattern led the poet, whether consciously or unconsciously, to assimilate the episode to the rest of that pattern.

It may, as we have just seen, be hard to determine whether a resemblance between two poems is due to direct borrowing or to common use of a tradition: and if direct borrowing seems likely, then the uncertain chronology of medieval epics raises the question of which poem was the borrower and which the source. Literary analysis of the passage concerned and of its context may provide the answer. If material appears either incongruous or superfluous in one poem whereas very similar material fits its context admirably in another poem, then the evidence for borrowing is strong. A classic case of such analysis is Martín de Riquer's study of the resemblances between the laments of Gonzalo Gústioz in the Cr1344 version of Siete infantes and Charlemagne in Roncesvalles³⁴. Menéndez Pidal pointed out some of these when he first published Roncesvalles in 1917, and they were subsequently discussed by other scholars. Opinion was equally divided between those who believed that a late refundición of Siete infantes borrowed from Roncesvalles, and those who held that the two poems made independent use of a common tradition. Siete infantes was not, until Riquer's investigation, taken seriously as the source for this part of Roncesvalles because of the general acceptance of Menéndez Pidal's view that the version prosified in Cr1344 had been composed in the fourteenth century (some decades after the composition of Roncesvalles). Riquer, however, approached the problem in a different way, noting that laments addressed to heads rather than corpses make perfect sense in Siete infantes but not in Roncesvalles, where the dead have not been decapitated: «Todo lo que parece forzado o innecesario en Roncesvalles es básico, obligado y consustancial en el Cantar de los Siete Infantes de Salas» (p. 211). Therefore a version of Siete infantes containing Gonzalo's full lament must have coexisted in the thirteenth century with the version prosified by EE, which disposes of the lament in a couple of lines. Cummins has recently shown (without reference to Riquer) that the poem used by EE must have continued to circulate in the fourteenth century, alongside the expanded refundición (see n. 5 above), and there is, as we shall see, good reason to extend even further backwards the coexistence demonstrated by Riquer.

A similar case occurs in the Mocedades de Rodrigo: the characterization of King Sancho Avarca as a huntsman, not long before his accidental discovery of the tomb of San Antolín at Palencia, seems to derive from the Poema de Fernán González, where the Count's discovery of the hermitage at Arlanza is the result of his hunting expedition³⁵. The superfluous element in Mocedades and the incongruous element in Roncesvalles are, in different ways, the result of conscious or unconscious borrowing from another epic. Material borrowed not from a single work but from a general tradition is much less likely to be conspicuous in this way, but it can on occasion be identified by its superfluity. Such is probably the case with the duels in the Cantar de Mio Cid. The villains of the poem are the two Infantes de Carrión, but the duels which round off their humiliation and the Cid's triumph are not two in number but three: Asur González, the Infantes' brother, who has hitherto made only a fleeting appearance (line 2172), involves himself in the arguments and challenges of the court scene, and is defeated by Muño Gustioz in the third duel. The poet makes good use of this addition, but there is no obvious structural or thematic need for it, and it is probably due to the symbolic power of the number three in folklore (strengthened, no doubt, by its Christian symbolism)³⁶.

One other case of this kind provides important evidence of a direct relationship between two poems of the Counts cycle. In the EE version of García, the Vela brothers:

ovieion ý su consejo malo et falso et de trayción de cómo matassen al inffant; et dixo Yénnego Vela: «yo sé en qué guisa podremos mover razón dond ayamos achaque por quel' matemos. Alcemos un tablado en medio de la rua, et los cavalleros castellanos, como son omnes que se precian desto, querrán ý venir a assolazarse, et nós bolveremos estonces pelea con ellos sobrell alançar, et matarlos emos a todos desta guisa».

(PCG, chap. 788, pp. 470b-471a)

Their later actions show that there was no question of attempting to make García's death appear semi-accidental. He comes on the scene only after the Castilian knights have been killed (the Alfonsine chroniclers contrast the account given in the Romanz with that of Lucas de Túy and Rodrigo Ximénez de Rada, who say that García was killed first), and the Vela brothers seize him, debate what to do with him, and then kill him. Moreover, their action in closing the city gates as soon as the quarrel begins, and then killing all the Castilian knights trapped within the walls, shows them for what they are -deliberate murderers. There is no practical advantage in engineering a quarrel, and we must look elsewhere for an explanation. It may be found in Siete infantes, where the chain of vengeance and counter-vengeance begins with just such a quarrel. As part of his wedding celebrations, «mandó Roy Blásquez parar un tablado muy alto en la glera cercal río, et fizo pregonar que quienquier quel' crebantasse quel' darié éll un don muy bueno» (PCG, chap. 736, p. 431b). From this harmless decision (such contests seem to have been a standard feature of festivities) come Doña Lambra's praise of Álvar Sanchez's skill, expressed in terms insulting to Gonzalo González and his brothers, the quarrel, and the fatal blow struck by Gonzalo. All this is a natural development; there is nothing forced or contrived about it. The incident of the tablado in García, contrived as the pointless pretext for a quarrel, seems to be modelled on it.

There are other similarities between the two poems. The characters of the two Sanchas are, as we have seen, transformed in the same way by the murder of their loved ones, and both of them, when asked to decide the fate of a murderer, choose a particularly cruel and bloodthirsty form of vengeance in which they will play an active part. It seems fairly clear that the incident of the tablado is borrowed by García from Siete infantes, but how much further does the debt go? The resemblance between the two Sanchas may well be too close to be acceptable as mere coincidence, and their identity of name and similarity of situation could easily have led to an almost total assimilation of character, as well as to the borrowing of the episode which starts the killings. García, unlike Siete infantes, has a firm historical basis for much of its plot, and the events concerned took place in 1029, several decades after the likely date of composition of Siete infantes. I suggest that, just as the historical reality of Sancho II's assassination seems to have been recast in a mythic pattern when it was given poetic form³⁷, so the reality of García's murder was poetically assimilated to the already established pattern of Siete infantes. There is, of course, one difficulty: the EE version of Siete infantes gives a much shorter and less bloodthirsty account of Mudarra's vengeance, in which Sancha is not involved (chap. 751, pp. 447-48). This obstacle is not, however, as serious as might at first sight appear. Riquer, as we have seen, makes a strong case for the existence in the thirteenth-century of a Siete infantes with features which are missing from the EE prosification but present in Cr1344, and it is possible that two or more versions, one of them with a fully developed vengeance involving Sancha, co-existed from a much earlier stage. It is even possible that the version used by the compilers of EE, though different from that of Cr1344 (as Menéndez Pidal demonstrated and Cummins has recently confirmed), was not in its poetic form as different as the chronicle texts suggest. The vengeance section in EE has, I have suggested above, a somewhat perfunctory air, which may indicate compression. It has, indeed, been argued by Angelo Monteverdi that all the differences between the story as given in EE and that in fourteenth-century chronicles can be explained by the varying treatment to which a single poem was subjected by the chroniclers³⁸. Other scholars have shown that this view is too extreme, but it may well contain an element of truth. But whether one concludes that the poetic version used by EE had a full vengeance narrative which the compilers abridged, or whether one prefers the theory of a poem with a fairly perfunctory vengeance co-existing with a much fuller narrative, there is no need to reject the literary evidence for García's dependence on a Siete infantes which began with the incident of the tablado and ended with vengeance exacted by Doña Sancha³⁹.

There are other similarities between poems in the Counts cycle, and between such poems and other epics, which deserve study. Some will no doubt prove to be due to independent use of folk motifs or of other traditions, and some to mere coincidence, but there may well be others which indicate direct relationships between the epics, and whose investigation would yield interesting results.

This epic, as prosified in EE, unquestionably belongs to the Cidian cycle. It is not merely that Rodrigo Díaz de Vivar dominates the action after Sancho's death. He plays a major part from the beginning, being entrusted by the dying King Fernando with the care of his sons and daughters: «Et allí mandó llamar a Roy Díaz el Cid que era ý, et comendó!' sus fijos et sus fijas que los consejasse bien et toviesse con ellos do mester les fuesse» (PCG, chap. 813, p. 494a). And these are not children, but grown men and women, soon to become rulers of cities and of kingdoms. Entwistle's view that, in the EE version of the Cantar de Sancho II, «The character of the Cid is taken from the Poema del Cid, where he already figured as somewhat greater than the emperor»⁴⁰, is one that most scholars would probably accept, even if some might jib at Entwistle's suggestion that the same poet may have been responsible for both epics. The resemblances go further than the Cid's stature in the two poems. Colin Smith notes that Sancho II is the only epic which compares with the Cantar de Mio Cid in its number of named Christian characters⁴¹, and Fraker, taking up this observation, maintains that these poems have in common sophistication of outlook and complexity of theme (p. 477), and respect for the law (pp. 484-86). Martínez, approaching the problem from another direction, finds noteworthy similarities in details of the two plots⁴².

It is clear, then, that the EE version of Sancho II shows the influence of the Cantar de Mio Cid, and it is probable that it comes from the same background. Its position in the Cidian cycle is, when it appears in EE, secure. Yet when we look at earlier evidence, a different picture emerges. The earliest evidence available is that of the Crónica Najerense which, it now seems clear, drew both on the Carmen de morte Sanctii regis and on a Cantar de Sancho II⁴³. Each could have been composed at any time between the events of 1072 and the date of ca 1160 which is generally attributed to the Najerense. The Carmen seems to have ended with Sancho's death and his burial at Oña, but the Cantar probably included the challenges and duels which followed it -the fictitious episode known as el reto a concejo.

Menéndez Pidal, Reig, and Martínez compare, in increasing detail, the versions of the story given in the Najerense and in EE, but only in Fraker's study do we have firmly contrasted outlines of two poems. Some details of Fraker's analysis may be questioned -I am not convinced, for example, that in asserting the manifest anxiety of the later poet to clear Urraca of complicity in her brother's murder (pp. 470-71), he gives enough weight to the well-known words which he quotes in a footnote: «Yo mugier só, et bien sabe él que yo non lidiaré con él, mas yol' faré matar a furto o a paladinas» (PCG, chap. 832, p. 507b). Nevertheless, the main lines of his argument (pp. 471-73) seem to me not only attractive but convincing. He sees the first Cantar de Sancho II as beginning with Fernando I's division of the kingdom among his sons and daughters, whereupon Sancho, as the eldest son, feels aggrieved that this abandonment of the rule of primogeniture has deprived him of his rightful inheritance. Sancho takes vengeance by invading the kingdoms of Galicia and León, which had been bequeathed to his brothers García and Alfonso, and he tries to complete his conquest by taking the city of Zamora, his sister Urraca's inheritance. Urraca, in desperate self-defence but also in revenge for the deposition and exile of Alfonso⁴⁴, arranges Sancho's murder by Bellido Adolfo. After the murder the Castilians seek revenge by the reto a consejo, and probably win the ensuing duels⁴⁵.

The first Cantar seems then to have been a story of vengeance and counter-vengeance, whereas the later poem, prosified in EE, weakened this chain of events by making García of Galicia (who had benefited greatly from his father's will) the first aggressor, and by making Urraca's connection with the murder ambiguous (or, if one takes Fraker's view, by clearing her entirely). It will not have escaped the reader's notice that emphasis on vengeance, and perhaps on counter-vengeance as well, is one of the common features of the Counts cycle. Another is, it will be recalled, the domination of part of the plot by a woman who is capable of violent action (and, in two out of the four epics of this cycle, of a treacherous murder). The first Cantar de Sancho II thus displays the two most significant common features of the Counts cycle. It has the other two as well -there is a historical element (one of considerable importance in this case), and the epic is connected with a tomb cult (that of Sancho's tomb at Oña). These two features are less distinctive and hence less significant, but the fact remains that, if we accept Fraker's analysis, the original Sancho II was strikingly similar to the four poems of the Counts cycle which have come down to us in prose or verse form. Fraker hints at this similarity when he refers to «the possibility of an early Cantar de Sancho which is much more conventional, much more like certain other Castilian epic songs, than is the version which survives in the Primera crónica» (p. 476). He does not refer specifically to the Counts cycle, nor does he single out the dominant woman as a feature, but these two points seem to me to be legitimate -indeed, inevitable- conclusions from the evidence he presents.

The later Sancho II, prosified in EE, has, as we have just seen, no chain of vengeance and counter-vengeance, and its blurring of Urraca's responsibility means that it has no dominant woman either. It still has a historical element and the connection with the Oña tomb cult, but the two most significant features linking Sancho II with the Counts cycle have gone. In their place is the Rodrigo Díaz of the Cantar de Mio Cid, who dominates the poem. The poetic tradition of Sancho and his assassination has, with this refundición, exchanged the features of one cycle for those of an unconnected poem -the only case of such a transfer that I am aware of in Spanish epic.

It is not hard to see why the first Cantar should have been cast in the mould of Siete infantes and the other poems of the Counts cycle. These may not have been the first Spanish epics (though the evidence for any earlier ones is weak), but they were in the eleventh century the only established epic cycle in Spain, and it would be natural for a new poem to follow the same pattern. This would be all the more natural if the events which gave rise to the new epic had some resemblance to incidents in the Counts cycle, as was the case with the events of Sancho's reign. The historical facts of his murder, together with what now seems to have been an incestuous, or at best an ambivalent, relation between Alfonso and Urraca, form a promising basis for a sensationalized epic plot analogous to those of the Counts cycle. What is more, Sancho was probably seen at the time (by some people, at least) as a successor to the Counts of Castile. A sense of continuity was strong: when Sancho el Mayor of Navarre divided his kingdom, his son Fernando, already proclaimed Count of Castile, received the western segment⁴⁶; and this was the Fernando who had married Sancha, fiancée of the murdered Count García. Their son Sancho, the future Sancho II, was born in 1032, only three years after García's murder, so it was easy enough to see him as the natural heir to the last Count of the native line.

By the mid-thirteenth century, however, the picture had changed. The epics of the Counts cycle no longer stood alone, but had a vigorous competitor: the Cantar de Mio Cid, which gave powerful and distinctive poetic form to the Cid's historical greatness⁴⁷. It is normal for the success of an epic about a hero's mature years to be followed by the composition of poems about earlier periods of his life, and here was material ready to hand: an epic in which the Cid already played a part, in which the chief character was the Cid's patron, and which could form a consecutive story with the Cantar de Mio Cid once the episode of the jura de Santa Gadea (a postscript to Sancho II, and a motive for Alfonso's hostility to the Cid in the first half of CMC) was added. Thus what had begun as a late appendage to one cycle became, in its new form, part of another.

Most of this article consists of speculation -my own, or the adopted speculations of others. This is inevitable, given the nature of the evidence (lost poems, chronicle accounts whose relation to the poems and to each other is uncertain, incomplete or reworked verse texts). I would make only two points in justification. First, in the absence of conclusive evidence, the only way of making progress is to speculate, while trying to ensure that the hypotheses offered are consistent with all of the available evidence, and that they are logically coherent. Secondly, most of what has previously been written about the development of Spanish epic is equally speculative, even if this is not openly admitted. Such work has generally been based on an investigation of the historical element in epic, and there may be some value in balancing it with speculations arising from a predominantly literary study. The picture of early Spanish epic which emerges is rather different, but it may be worth consideration.

It is customary in neotraditionalist criticism to assert that the original versions of Siete infantes, García, and the rest were composed soon after the events, were only a few hundred lines long, were historically accurate, had no connection with tomb cults or other ecclesiastical interests, and lacked the familiar elements of traditional fiction (the claim that the early epics lacked dominant, bloodthirsty women is part of this doctrine). Now, it may sometimes be necessary to conclude that a particular extant text, such as the Cantar de Mio Cid, has such strongly individual characteristics that it is probably unlike any of its predecessors, but it is surely imprudent to assert that all the versions of a group of poems, as far back as we can trace them, are, in all their most obvious characteristics, quite unlike the original poems of that group. We know nothing about the lost originals by direct observation, and our only basis for forming conclusions about them is a study of the surviving material. That study must, of course, include a comparison with what can be independently established about the historical background, but it must also include a literary comparison of the Spanish epic material with, its analogues in other languages and other genres. The conclusion to be drawn is, I think, that the lost originals were probably rather similar to the material surviving in verse texts and chronicle prosifications. There is certainly no need to assume that fictions of the type embodied in the Counts cycle need a long time to evolve. I see no reason why we should not apply to these poems a view expressed in another context by Menéndez Pidal himself. Arguing against the belief of Milá y Fontanals and Menéndez y Pelayo that the epic of Mainet could hardly have been based on Alfonso VI's exile in Toledo, and that «es muy poco... medio siglo para que se cree una leyenda que transforma a Alfonso VI en Carlomagno», Don Ramón wrote: «Es que estos críticos suponen, según las ideas románticas, una lenta gestación de toda leyenda en el alma del pueblo, sin admitir la caprichosa invención de un juglar que, al recordar las aventuras de Alfonso, fantasea las de Carlos»⁴⁸.