Collection, composition, and compilation in the Cantigas de Santa Maria.

Abstract. Accounts of the production of the Cantigas de Santa Maria assume that the rich manuscripts were produced by poets, artists, and musicians working together. This paper develops the complexities of three central phases or processes--collection (the acquisition of narrative materials, and their preparation for use), composition (the separate production of text, music and images), and compilation (the assembly of structured sequences in the different manuscripts)--and their complex interactions. The traditional question of sources can be profitably redefined as relevant to all three phases. Special attention is given to the extent to which the different compilation phases retain or disperse groups of stories collected from the same source, and to the processes of recomposition which take place when poems are revised in the process of compilation or recompilation.

Keywords. Cantigas; miracles; manuscripts; poetry; music; sources

Resumo. No estudo das Cantigas de Santa Maria prevalece a visao simplificadora de poetas, artistas e musicos a trabalharem em conjunto para elaborar os ricos codices de Afonso X. A presente abordagem considera a complexidade e a interaccao dos tres processos centrais de coleccao (aquisicao de materiais narrativos e a preparacao destes para exploracao), composicao (elaboracao de texto, musica e narrativa grafica) e compilacao (ordenacao e incorporacao nas sequencias altamente estruturadas dos codices) propondo uma reinterpretacao da velha questao das fontes das CSM. Nas diversas fases de compilacao e recompilacao os agrupamentos de cantigas coleccionadas da mesma fonte hagiografica ora se mantem, ora se dispersam. O processo de (re)compilacao pode obrigar a recomposicoes.

Palavras-Chave. Cantigas; milagres; manuscritos; poesia; musica; fontes

**********

The Cantigas de Santa Maria (hereafter CSM), (1) the great work of Marian devotion masterminded by Alfonso X of Castile, combines and coordinates three artistic enterprises--a versified miracle collection, a vast repertoire of monodic song, and a wealth of illustrative miniatures. The frequently reproduced presentational images which precede the main body of cantigas in MSS E and T provide a seductive picture of uniformity and simplicity, in which the King, central to all processes, directs the production of Music, Text and Images, with the implication that the combination of these three components of what Keller labels the 'threefold impact' was just a matter of a well-ordered production line of scribes and artists. (2) The reality is much more complex, and its complexity is important not only for our understanding of the literary and cultural context of the CSM, but also for the processes of editing and commentary to which the CSM have yet to be systematically subjected.

It is essential to recognize that the Alfonsine project had at least three 'editions', differing not only in their dimensions but in their ethos and construction. The initial compilation of a hundred in To is primarily an exercise in poetic devotion, symmetrically structured in two blocks of fifty cantigas, and with appendices showing its potential for expansion. To is transformed through progressive expansions (first to 200 and then to 400) into the monumental decorated compilation of T/F, in which the element of graphic decoration is primary. Finally E (the largest of the compilations, comprising 400 cantigas with prologue, epilogue and festal cantigas, but in no sense Alfonso's definitive version) represents not the culmination of the enterprise but the endgame by which a complete set of cantigas is recorded to back up the unsustainable effort of T/F. (3) Despite returning to the undecorated format of To, E retains features of T/F which no longer serve any useful purpose, such as the association of the number 5 with long cantigas; it also contains the distinctive group of Puerto de Santa Maria poems, the last nucleus of shrine-poems to be incorporated. At the same time, it must be recognized that individual cantigas have an independent textual history which does not reduce to the histories of the MSS in which they are recorded.

To begin to understand the complexity, we need to divide the production process into three active phases: collection, composition and compilation. (4)

Collection is the process of acquiring narrative or literary materials. This should not be seen as a passive process of reception but an active one of procurement (or, in modern commercial parlance, 'sourcing'). It includes the incorporation of ready-made miracle stories in written or oral form, the borrowing of schematic tales from secular or religious traditions, and the recording of personal recollections. In the case of tales extant in foreign languages (Latin and French, at least), the process of collection involves (or implies as a subsequent stage) translation from the source language, and/or summarization in a usable form. A similar intermediate process must be assumed as part of the translation of narratives from a textual form into the visual narratives of the miniatures of T and F.

Composition is the process of production of narratives. Here it must be recognized that there are three compositional processes involved, resulting in texts, miniatures, and music respectively. It is too lightly assumed that even if the composition of the texts was prior to the composition of the accompanying images and music, they were closely coordinated or nearly simultaneous. This is implicit in much work stimulated by the publication of the facsimile of T, which makes the assumption that the miniatures can clarify points of textual obscurity. (5) Snow, assuming royal authorship for most of the collection, draws attention to the passages such as 'E dest' un miragre, de que fiz cobras e son' (64.6) in which it is asserted that text (cobras) and music (son) were created together. (6) The degree to which these assumptions are true, and the extent of the intermediate processes of summary, translation, and visualization of narratives to enable the production of images, will be part of the individuality of a cantiga's history.

Finally, compilation is the assembly of the component narratives into the ordered and structured sequences found in the MSS. (7) It is not just a question of the sequence of copying of individual pieces, but includes the processes by which words and music were brought together, the mechanics of the layout of text, music and images, the incorporation of additional textual material such as epigraphs and captions, the construction of indexes, and the formation of appendices or satellite compilations (as found in To and E). The processes of compilation may indeed lead to composition (of the epigraphs preceding each cantiga, index entries, captions to individual panels in the miniatures), recomposition (when a cantiga needs to be modified for inclusion, or more than one cantiga is extracted from a single source), emendation (when text and music are found not to match), duplication (repetition of pieces to fill gaps in compilation programme, as is found at the end of E), and exclusion (of pieces for which no place can be found). (8) Examples of many of these will be discussed in the body of this article.

From this more complex view of the constituent processes of the production of the CSM compilations, the team of participants working under the King would have included miracle collectors, summarizers and translators; archivists and filing clerks maintaining the central repository of stories; composers of texts, music and visual narratives; compilers and indexers of completed poems; and organizers of manuscripts, directing copyists in the scriptorium or workshops.

One enigmatic cantiga provides an example of the usefulness of the distinction of these phases, particularly compilation and composition. presents a number of apparently unrelated oddities. (9) Firstly, it is separated from the three other Montserrat miracles (To 72, 62, 66) with which it can be assumed to have entered the archive, as its position at the beginning of the second hundred suggests it was present early. Secondly, its first stanza shows notable hesitation in metrics, with the music scribe of E making various amendments to make the text fit the music. Thirdly, in T it has no music at all. Fourthly, its miniatures show a curious correlation with the text, in that the first two of the six illustrate the refrain, not the narrative, and the remaining four show an order of events different from the text. The internal evidence of text, metrics and punctuation points to the T and E versions being an expanded, rewritten version of a very short narrative, allowing us to reconstruct the following extended sequence of stages of production: the story would have been collected with other Montserrat stories, and a first poetic version composed, with music, but with metrical ambiguities complicating the fit between music and text; this in turn led to its not being incorporated in the first compilation; once included in the body of poems for inclusion in T, its shortness (too short to occupy a page in T) led to its recomposition as a longer piece, which came too late to change the work of the illustrators (graphic composition) who had been give instructions based on the short version, and did not solve the metrical problems sufficiently for the music scribe to complete the compilation by adding music to the blank staves; finally the expanded poem was passed to the team compiling E, whose music scribe then amended the text.

This separation of phases needs to be carried into the story of the compilations. It is clear that the expansion of the compilations from To to T/F entailed a complete revision both of the programme of collection (from a reliance on 'international' miracles using multiple standard sources, to the systematic exploitation of local miracles using written or oral sources, and recent historical events) and of the principles of compilation (from the simple structuring of To into ten decades, to the layout-driven order of T/F, and finally the pragmatic forced completion of E), leading to the wholesale revision of the order of presentation of the poems which had already been used in To. There is evidence for a slower change in compositional practices, with the retreat from metrical virtuosity reflected in the less ambitious verse forms which predominate in the last 200 poems. (10)

This change in orientation has been--and continues to be--utterly obscured by the use of E/Mettman numbers as the only reference scheme for the CSM, which has hidden the evidence provided by To for the deployment of sources of the first compilation, and has left the reconstruction of the order of F an exercise of purely academic interest. (11) We cannot automatically equate 'the first hundred cantigas' with the body of To, as 10 per cent of To (10 cantigas) was either discarded or moved to later in the compilations. It is an interesting fact--but a different fact--that the balance of local and international tales is the same in these two 'first hundreds', which reflects Schaffer's observation that the main shift in sources did not come until after the first 200 poems had been assembled. (12)

This new perspective also leads us to deconstruct the issue of the 'sources' of the CSM. We can distinguish three types of source, corresponding to the three phases we have identified: archive (collection-source), text (composition-source), and model (compilation-source). For each phase we need to ask different questions.

On the collection phase, as well as investigating the origins of the individual miracle stories which Alfonso's poets versified, we need to ask whether he used one single precursor (international) collection such as the much-discussed mariale magnum, a reference library of (international) collections, local written and oral collections, or his own personal archive, of tales drawn from a wider range of collections (including historical and personal miracles to complement those drawn from the hagiographical tradition). (13) Assuming the latter, how did this archive evolve in the expansion from 100 to 400?

On the composition phase, the questions concern the extent to which the text of the CSM (narrative structure, narrative content, phraseology) can be traced to particular (or multiple) source texts. (14) In the cases where the same narrative recurs, to what extent do the different manifestations of the narrative use the same source? (15)

Focusing on compilation, we investigate the extent to which the order and organization of poems in the CSM reflects their order of presentation in source or precursor compilations. Did any of these supply a model for the CSM as miracle compilation? Possible answers have included the identification of Gautier de Coinci's Miracles de Nostre Dame as a model both for a cancioneiro of versified miracle stories and for the structured interlacing of lyrical and narrative poems; the psalms, with their complement of 150, as a model for an intermediate stage of compilation; and the idea that the mariale magnum itself could have been a model or example to be outdone. (16) This way of posing the question also obliges us to give To its due precedence as the first compilation, and indeed the only compilation which was assembled without reference to any of the others. Remarkably little attention has been given to the question of how To itself was originally organized, as most scholars have preferred to address the question of the reorganization of its contents, emphasizing the importance of the quint principle by which long cantigas in T/F were placed in the structurally prominent position of the mid-point--the fifth cantiga--of every decade.

A telling example is provided by cantigas 35 and 362, which concern a wonder-working reliquary of the French cathedral of Laon. They are based on stories first told c. 1150 by Herman, a canon of the cathedral. (17) In a typical confusion of names, the texts of these cantigas place them in Leon do Rodao, alias Lyon du Rhone, rather than Laon, but it is clear that Laon is intended. More importantly, the narration of cantiga 362 refers back to cantiga 35 in an allusive way:

   Esta foi aquela arca | de que vos eu ja falei
   que tragian pelo mundo | por gaar, segund' achei
   escrito, porque ss' a vila | queimara, como contei
   outrossi, e a ygreja | toda senon o altar (362. 19-22)

A few lines later, the narrator alludes to the relics contained in the chest: 'E grandes miragres fez por elas Santa Maria, como vos dix' outra vez.' (29-30). In MS E, these narratives are separated by 326 others, so a reader would have to be very attentive indeed to understand these cross-references. (18) However, in To, only two miracles intervene between these cantigas, as CSM 35=To 92 and CSM 362=To 95, (19) and they would therefore have been easily connected in the minds of their readers. In fact, E is the source in which these two miracles are furthest apart: in T/F, the first cantiga is no. 35 in T and the second is no. 42 in F. The initial reason for the separation of the two cantigas is the quint principle of T/F: To 92 was a suitable candidate for quint, being a particularly long song, lending itself to two full-page miniatures instead of just one, and its promotion to the fourth decade of T (T35) separated it from the other Laon miracle. Not only does the quint principle explain the order in T, but the original order in To shows a degree of clustering of miracles from the same location in the first compilation. (20)

The textual and structural source of these cantigas could be either Herman's original Latin account or Gautier de Coinci's verse redaction based on it. (21) The two miracles follow immediately after one another in both compilations (To 92=Herman II.4 /Gautier Bk II, 15 and To 95=Herman II.2 / Gautier Bk II, 14). But we still need to explain why these cantigas are incorporated into To in the opposite order from the one in which they occur in Herman and Gautier, which is also the historical order of the events concerned (the reliquary was taken first to N. France and only then to Britain). This is done quite knowingly, as the narrative of To 92=CSM 35 is expanded to provide the details of the journey, which in the originals is provided by reading the miracles in order:

   E andou primeiro Franca | segundo com' aprendi
   u fez Deus muitos miragres | per elas; e foy assy
   que depois a Ingraterra | ar passou e com' oy
   polas levar mais en salvo | foy-as na nave meter. (35.35-38).

One should also ask why the two stories from the same source were not placed immediately one after the other in To. It seems that the compilers of the first compilation were happy to disguise the extent of their dependence on known sources. Small nuclei of stories are interspersed with stories from other locations: the first small group of miracles from Montserrat (To 62, To 66, To 72) is broken up by other tales, and in two cases followed by a Soissons miracle (To 63, 67). The location itself can be downplayed or altered. To 95, while forming part of the Laon cycle, is identified in its epigraphs as a miracle located in Chartres 'Como Santa Maria fez cobrar seu lume a un ourivez en Chartes'. (22) The Montserrat miracles, in the later versions, lose their geographical identity by being apparently translated to Monsaraz in Portugal. (23) A case of mislocation by association is found in E 82=To Appendix V, set in Canterbury in the CSM and in an unnamed Carthusian monastery in its textual source, Vincent de Beauvais. The explanation seems to lie in the fact that the miracles flanking it in Vincent (used as E85 and E288) are both set in England, and the last in Canterbury itself. (24)

An example of a closer grouping of poems from a single source is provided by the French local miracles, first recorded by Hugo Farsitus, most concerning the healing of people suffering from S. Martial's Fire in Soissons in 1128. (25) If we use E numbers, these miracles are well dispersed. (26) But if we consider their arrangement in To, we can detect a sequence of six stories (To 44-To 49) forming a cluster. (27) Clearly they were incorporated as a block, and were later dispersed: but from where did they originally come? To begin, one must consider whether Alfonso X drew these miracles directly from the Latin text of Hugo Farsitus or received them indirectly from the works of another thirteenth-century compiler, such as Vincent of Beauvais; (28) Gautier de Coinci; or Juan Gil de Zamora. (29) There are thirty-one miracles in Hugo's book; twelve of these occur in the CSM, but only four each in the compilations of Gautier de Coinci and Vincent de Beauvais. Gil de Zamora incorporates eighteen in his Liber Mariae, but one of the Soissons narratives occurring in the CSM does not appear in his work. It is clear, therefore, that Alfonso X had recourse to a more comprehensive source. According to Mettmann the six cantigas which form a cluster in To are definitely based on Hugo's text. (30) We do not, however, need to conclude that an exemplar of Hugo's book was necessarily the direct textual source of To 44-49, because its text (and order) is accurately preserved in later compilations. It is very likely that Alfonso X's compilers consulted Hugo's work in one of the Marialia which include, in distinct blocks and with little textual modification, French shrine miracles of Laon, Soissons, Chartres and Rocamadour. (31) Marialia were undoubtedly used by Vincent of Beauvais, when writing his miracles of the Virgin in the Speculum historiale, by Gil de Zamora, who took Soissons miracles from a Mariale and presented them in his Liber Mariae in clusters, maintaining Hugo's original order, and by Gautier de Coinci, who claimed to base his work on a book he found in his own monastery of St. Medard. (32) Could the compilers of the CSM also have consulted such a volume?

Figure 1 shows that in neither To nor T is there any correlation between the order of the CSM and the order of these Soissons narratives in Farsitus or in any of the other compilations linked to the CSM, so the structural and archival source must be presumed to be a Soissons cluster created by the scriptorium. Like the Laon miracles, their original location is not always identified. To explain their dispersal we need a fuller picture of the planning of T/F.

It is clear that the compilers of T/F took the Toledo compilation as the basis for their work, as the first hundred of T/F are clearly a reordered version of To rather than a new selection. The reordering followed a clear pattern. The compilers first took the loores more or less intact (except for To 50 and 100, loores structuring the first compilation, and thus incompatible with the new), and in their original order, to make the decadal structure of the new compilation. They then extracted two groups of existing miragres: potential quints (nos 7, 12, 17, 19, 26, 33, 38, 55, 81, 83, 86, 88, 92, 97 99), which would provide the complementary structure to the loores; and cantigas of problematic content or layout (21, 28, 31, 42, 45, 46, 48, 53, 74, 76, 77, 79, 84). Quints were re-inserted into the sequence, including some new ones, and the remaining cantigas (including some potential quints not used as such) were fitted in more or less in their original order, with some adjustment to get the right number of pages between successive quints (and later the right number of pages to allow miniatures to be placed on a facing recto). (33) Figure 2 shows, in summary form, how the redistribution of the first fifty cantigas involves relatively little change of order in those retained and not redesignated as quints. Figure 3 shows the same information in graphic form, with the T compilation as the central spine and the To cantigas arranged around it. (34) Some (notably 47, 49, 51, 52) are held over and moved into vacant positions in the second half (61-64). The last thirty cantigas, by contrast, are in a more random order, with some of the problem poems incorporated in nos 80-100, and a significant number of cantigas are held back until the first decades of the second hundred.

[FIGURE 3 OMITTED]

As a result of all these redistribution processes, we can chart the dissolution of the Soissons cluster of 44-49. The six poems are divided into two groups: the unproblematic 44, 47 and 49 retained in the first 60 (44 in situ, and 47/49 slightly later) while the problem cases 45, 46, 48 are held over.

One factor in the identification of problem cases is clearly related, though at a distance, to layout. To 21 reappears as CSM 87, which was long ago identified as metrically highly problematic. (35) This is a cantiga in which the refrain is variable (constructed in parallel like a cantiga de amigo) so that the opening refrain is repeated after the first, but only the first stanza.

Original version of cantiga 87 (stanzas 1-2):

   Muito punna dos seus onrrar
   sempre Santa Maria

   E desto vos quero contar
   un gran miragre que mostrar
   quis a virgen que non a par
   na cidad de Pavia.
   Muito punna dos seus onrrar
   sempre Santa Maria

   Un crerig' ouv' i sabedor
   de todo ben e servidor
   desta groriosa sennor
   quant' ele mais podia.
   D'onrrar os seus a gran sabor
   sempre Santa Maria

However, the metrical problems are not obvious in To, which does no more than repeat the original refrain after the varied refrain. T does nothing different. Only in E does the structure go mad and generate a complex repetition, requiring a complex layout:

To 21                           E 87

Muito punna dos seus onrrar     Muito punna dos seus onrrar
sempre Santa Maria              sempre Santa Maria

E desto vos quero contar        E desto vos quero contar
un gran miragre que mostrar     un gran miragre que mostrar
quis a virgen que non a par     quis a virgen que non a par
na cidad de Pavia.              na cidad de Pavia.
Muito punna dos seus onrrar     Muito punna dos seus onrrar
sempre Santa Maria              sempre Santa Maria

Un crerig' ouv' i sabedor       Un crerig' ouv' i sabedor
de todo ben e servidor          de todo ben e servidor
desta groriosa sennor           desta groriosa sennor
quant' ele mais podia.          quant' ele mais podia.
D'onrrar os seus a gran sabor   Muito punna dos seus onrrar
sempre Santa Maria              sempre Santa Maria
Muito punna dos seus onrrar     D'onrrar os seus a gran sabor
sempre Santa Maria              sempre Santa Maria
                                Muito punna dos seus onrrar
                                sempre Santa Maria

This would suggest that what was consulted to decide where to put the cantiga was not the copy in To, but the exemplar in the archive. T hesitated, and put it aside for later, before making the same small mistake, and E attempted to interpret the original without reference to the other compilations, with disastrous results.

STEPHEN PARKINSON AND DEIRDRE JACKSON

CENTRE FOR THE STUDY OF THE CANTIGAS DE SANTA MARIA

Oxford University

(1) This work was supported by Leverhulme Trust project grant F/08 736/B 'Collection, Composition, and Compilation in the Cantigas de Santa Maria' (Project Director: Dr Stephen Parkinson; Research Assistant: Dr Deirdre Jackson), leading to the creation of the Cantigas de Santa Maria Database (http://csm.mml.ox.ac.uk) in the Centre for the Study of the Cantigas de Santa Maria of Oxford University. Earlier versions of parts of this paper were presented at the 16th Colloquium of the Medieval Hispanic Research Seminar, QMUL, June 2004, under the title 'Alfonso's Mariale Magnum', and at the Coloquio Universitario Internacional Cantigas de Santa Maria: Textos e Contextos, Santiago de Compostela, July 2004, under the title 'Coleccao, composicao e compilacao nas Cantigas de Santa Maria'. The conventional sigla for the CSM manuscripts used here are: To = Madrid, Biblioteca Nacional, MS 10069 (reproduced in Alfonso X O Sabio, Cantigas de Santa Maria, Edicion facsimile do codice de Toledo (Santiago de Compostela: Consello da Cultura Galega, 2003); E = Real Monasterio de S. Lorenzo de El Escorial, B.I.2 (codice de los musicos), reproduced in Higinio Angles (=Higini Angles), La musica de las Cantigas de Santa Maria del Rey Alfonso el Sabio, vol. 1 (Barcelona: Biblioteca Central, 1964); T = Real Monasterio de S. Lorenzo de El Escorial, T.I.1 (codice rico), reproduced in Cantigas de Santa Maria: Edicion facsimil del Codice Rico T.I.1 de la Biblioteca de San Lorenzo el Real de El Escorial, Siglo XIII (Madrid: Edilan, 1979); F = Florence, Biblioteca Nazionale, Banco Rari, 20, reproduced in Alfonso X el Sabio, Cantigas de Santa Maria, Edicion facsimil del codice B.R.20 de la Biblioteca Nazionale Centrale de Florencia, Siglo XIII (Madrid: Edilan, 1989).

(2) Martha E. Schaffer, 'The "Evolution" of the Cantigas de Santa Maria: the Relationships between Manuscripts T, F, and E', in Cobras e Son: Papers on the Text, Music, and Manuscripts of the 'Cantigas de Santa Maria', ed. by Stephen Parkinson (Oxford: Legenda, 2000), pp. 133-53, reproduces the images (196-99). John Esten Keller, 'The Threefold Impact of the Cantigas de Santa Maria: Visual, Verbal, and Musical', in Studies on the 'Cantigas de Santa Maria': Art, Music, and Poetry, ed. by Israel J. Katz and John Esten Keller (Madison: Hispanic Seminary of Medieval Studies, 1987), pp. 7-33.

(3) Martha E. Schaffer, 'Los manuscritos de las CSM: su problematica', in El Scriptorium alfonsi: de los Libros de Astrologia a las 'Cantigas de Santa Maria', ed. by Jesus Montoya Martinez and Ana Dominguez Rodriguez (Madrid, Editorial Complutense, 1999), pp. 127-48; Schaffer, 'Evolution', p. 189; Stephen Parkinson, 'The First Reorganisation of the Cantigas de Santa Maria', Cantigueiros, 1.2 (1988) 91-97; 'Layout in the Codices Ricos of the Cantigas de Santa Maria', Hispanic Research Journal, 1 (2000) 243-74.

(4) To make the distinction clearer, I refer to the different 'editions' of the CSM as 'compilations', rather than 'collections'.

(5) Connie L. Scarborough, 'Verbalization and Visualization in MS T.I.1 of the Cantigas de Santa Maria: The Theme of the Runaway Nun', in Studies on the Cantigas de Santa Maria (see Keller, above), pp. 135-54.

(6) Joseph Snow, 'Self-conscious Reference and the Organic Narrative Pattern of the Cantigas de Santa Maria of Alfonso X', in Medieval, Renaissance and Folklore Studies in Honor of John Esten Keller, ed. by Joseph R. Jones (Newark: Juan de la Cuesta, 1980), pp. 53-66.

(7) Schaffer, 'Evolution', p. 187, acutely observes that decision-making and execution must be distinguished as aspects of compilation: 'the compilation of the collections in some cases preceded the moment at which the codex in which they are preserved was designed and/or executed'.

(8) Martha Schaffer, ' "Ben vennas mayo": A "failed" cantiga de Santa Maria', in Estudos Galegos Medievais, ed. by Antonio Cortijo Ocana, Giorgio Perissinotto, and Harvey L. Sharrer (Centro de Estudos Galegos: University of California Santa, Barbara, 2001), 97-132, explains the exclusion of CSM 406. Parkinson, 'Reorganisation', discusses the exclusion of CSM 50.

(9) See Stephen Parkinson, 'The Evolution of Cantiga 113: Composition, Recomposition, and Emendation in the Cantigas de Santa Maria', to appear in La coronica, for an extended discussion of this complex case. John Esten Keller, 'Montserrat in the Cantigas de Santa Maria of Alfonso X, el Sabio', in Josep Maria Sola-Sole: Homage, Homenaje, Homenatge, ed. by Antonio Torres-Alcala (Barcelona: Puvill, 1984), pp. 67-78, describes the miniature from the narrative point of view.

(10) Parkinson, 'Phonology and Metrics: Aspects of Rhyme in the Cantigas de Santa Maria', in Proceedings of the Tenth Colloquium, ed. by Alan Dyermond (PMHRS 30; London: Queen Mary and Westfield College, 2000), pp. 131-44 (p. 138).

(11) Nella Aita, O Codice Florentino das Cantigas do Rey Affonso, o Sabio (Rio de Janeiro: Litho-Typ. Fluminense, 1922); Deirdre Jackson, 'A Proposed Reconstruction of the Disordered Quires of Florence, Biblioteca Nazionale Centrale, Banco Rari 20', in Conference Proceedings: The City and the Book, International Congress, Florence 4-6 September 2002 <http://www.florin.ms/beth2.html>.

(12) Schaffer, 'Evolution', p. 190.

(13) Adolf Mussafia, 'Studien zu den mittelalterlichen Marienlegenden', Sitzungsberichte der kaiserl. Akademie der Wissenschaften in Wien, philosophisch-historische Classe, 113 (1886), 917-94; 115 (1888), 5-92; 119, no.9 (1889), 1-66; 123, no.8 (1891), 1-85 and 139, no. 8 (1898), 1-74; Paule V. Beterous, Les Collections des Miracles de la Vierge en Gallo et Ibero-Roman au XIIIe Siecle. Marian Library Series. New Series, Vols. 15-16 (Dayton, OH: University of Dayton, 1983-84).

(14) For the concurrent sources of cantiga 7, Stephen Parkinson and Deirdre Jackson, 'Putting the Cantigas in Context: Tracing the Sources of Alfonso X's Cantigas de Santa Maria', paper presented at the International Medieval Congress, Kalamazoo, May 2005.

(15) Cantigas 223 and 225, for instance, give variants of a folktale in which a priest swallows a spider which has fallen into the communion chalice. No immediate source has been identified for either.

(16) As Cantigas de Loor de Santa Maria (edicion e comentario), coord. by Elvira Fidalgo (Santiago de Compostela: Xunta de Galicia, 2003), p. 22: 'Cando o rei Afonso decide componer un cancioneiro na honra da Virxe, o modelo mais afin ao que sen dubida recorreu foi a obra de Gautier de Coinci'; David Wulstan, 'The Compilation of the Cantigas of Alfonso el Sabio', in Cobras e Son pp. 154-85 (p. 169), cites Gil de Zamora's reference to the psalms, 'more quoque davitico etiam [ad] preconiam virginis gloriose multas et perpulchras composuit cantilenas', as evidence for a compilation of 150; Stephen Parkinson and Deirdre Jackson, 'Alfonso's Mariale Magnum', paper presented at the 16th Colloquium of the Medieval Hispanic Research Seminar, QMUL 2004.

(17) Hermanni Monachi, De Miraculis S. Mariae Laudunensis de Gestis Venerabilis Bartholomaei Episcopi et S. Nortberti, Patrologia Latina, vol. 156, cols. 975-82.

(18) But see Schaffer 'Evolution', p. 191, on the frequency of back references in CSM 360-400.

(19) For charts of the correspondences between the cantigas in the four MSS, see Walter Mettmann (ed.), Alfonso X el Sabio, Cantigas de Santa Maria, Clasicos Castalia (Madrid: Castalia, 1986-89), vol. I, 35-40, and Wulstan 'Compilation', pp.155-63.

(20) Philip Vandrey, 'Canticles 35 and 362: Keys to the Poetic Style of the Cantigas de Santa Maria', Kentucky Romance Quarterly, 27 (1980) 445-58, suggests that the two poems are by the same composer, on grounds of geographical origin and some unsurprising coincidences of metrical pattern.

(21) Elise Forsythe Dexter, 'Sources of the Cantigas of Alfonso el Sabio', PhD thesis (University of Wisconsin, 1926), Table I. Teresa Marullo, 'Osservazioni sulle Cantigas di Alfonso X e sui Miracles di Gautier de Coincy', Archivum Romanicum, 18 (1934), 495-539 (p. 501). In Herman's libellus these miracles occur in chapters II and IV respectively. In his discussion of these narratives, Beterous, Collections, p. 386, relying exclusively on E, wrongly states that Alfonso follows neither the order of Herman nor that of Gautier.

(22) Mettmann, Cantigas, III, 234.

(23) Parkinson, 'Cantiga 113'. The Cantigas de Santa Maria Database, <http://csm.mml.ox.ac.uk>, contains details of sources.

(24) Deirdre Jackson, 'Tradition and Innovation in Cantiga 82: the case of the demon swine', to appear in Cantigas United: Papers on the Galician-Portuguese Lyric, ed. by Stephen Parkinson, PMHRS.

(25) Hugonis Farsiti, Libellus de Miraculis B. Mariae Virginis in Urbe Suessionensi, J.-P. Migne, Patrologiae cursus completus, Series Latina, 179, col. 1176-1800.

(26) Evelyn Faye Wilson, The Stella Maris of John of Garland Edited, Together with a Study of Certain Collections of Mary Legends Made in Northern France in the Twelfth and Thirteenth Centuries (Cambridge, MA, 1946), 69, observes that some of Hugo's miracles occur in the CSM 'scattered throughout the entire collection'.

(27) Walter Mettmann, 'Die Soissons-Wunder in den Cantigas de Santa Maria', in Homenagem a Joseph M. Piel, ed. by Dieter Kremer (Tubingen: Niemeyer, 1988), pp. 615-20 (p. 616).

(28) As mentioned in his will, Alfonso X owned a copy of Vincent's work commissioned by King Louis IX himself. Georges Daumet, 'Les testaments d'Alphonse X le Savant, roi de Castille', Bibliotheque de l'Ecole des Chartes, 67 (1906), 70-99 (p. 90).

(29) Gil de Zamora dedicated an Office of the Virgin (Officium Almifluae Virginis) to the King, and appended it to his Liber Mariae.

(30) Mettmann, 'Soissons-Wunder', p. 617. 'Bei den sechs Gedichten steht fest, dass der Autor das Buch von Farsitus vor sich liegen hatte.'

(31) Benedicta Ward, Miracles and the Medieval Mind: Theory, Record and Event, 1000-1215 (London: Scolar Press, 1982), 133-34.

(32) Mussafia, 'Studien', 119, no.9 (1889), 1-66 at 34, asserts that one of two sources used by Gil de Zamora was 'eine der franzosischen Sammlungen', the other probably was the Golden Legend.

(33) Parkinson, 'Layout', pp. 259-62; Schaffer, 'Evolution', p. 194.

(34) A full colour version of Figure 3 is viewable on the website of the Centre for Studies of the Cantigas de Santa Maria (see note 1).

(35) Stephen Parkinson, 'False Refrains in the Cantigas de Santa Maria', Portuguese Studies, 3 (1987) 21-55.

Fig. 1: Some Farsitus miracles in CSM and its possible sources

To   T (E)   Farsitus   Gil de Zamora   Gautier de   Vincent de
                        tr. 16          Coinci       Beauvais

44   41      17         4, 15           --           --
45   106     25         3, 1            --           --
46   101     11         4, 10           --           --
47   61      12         4, 11           II, 72       --
48   81      7          4, 7            II, 73       27, 3
49   62      26         3, 2            --           --
63   49      18         5, 17           --           --
67   53      9          4, 8            II, 71       27, 4A
82   91      1 & 2      --              --           27, 2

FIG. 2: The reordering of To

a) Loores from To to T

   To 10 20 30 40 60 70 90 80
    T 10 20 30 40 50 60 70 80 90 100

b) Quints from To to T

   To 33 19 38 92 83 86 88 99        81  55  97
    T  5 15 25 35 45 55 65 75 85 95 105 115 125

c) Relocation of unused potential quints

   To  7 12 17 26
    T 17 16 24 21

d) Problematic cantigas removed from the main sequence of To

   To 21  28 31   42   45  46 48 50 53   74
    T 87  86 94   79  106 101 81 -- 78 (255)
   To 76  77 79   84  100
    T -- 132 -- (317)  --

e) Regular passage of the remaining poems from To order to T order
(location of quints and loores marked by Q and L)

   To 1-4   5-6 8-9   11 13-15   16 18   22-23
    T 1-4 Q 6-7 8-9 L 11 12-14 Q 18 19 L 22-23 Q

   To 24-25 27 29   32 34-6   37 39 41 43   44
    T 26-27 28 29 L 31 32-4 Q 36 37 38 39 L 41

   To 57 56 58   59 61-63   64 66 67 69
    T 42-43 44 Q 46 47-49 L 51 52 53 54

   To   47 49 51-52
    T L 61 62 63-64

f) Re-orderings from To to T (problematic cantigas are italicized)

   To   78 65 68 54   91 89 87   53# 42#
    T Q 66 67 68 69 L 71 73 74 Q 78  79  L

   To 48# 98   28# 21#   82 85 31#   94  46#  93
    T 81  84 Q 86  87  L 91 92 94  Q 98 101  103

   To  96  45#  77
    T 104 106  132

Note: Problematic cantigas is indicated with #.