Attempts to relate V’s musical offerings to those of other chansonniers have stumbled over rigid juxtapositions: orality is set up in contrast to written tradition, central sources are contrasted to marginal, while unreliable scribes are to be weeded out from the more reliable. V’s case shows, in microcosm, the problems of genetic research as applied to the transmission of melody in this repertoire. In the 19th century, the source was interpreted as a ‘central’ source in the stemmatic work of Eduard Schwan, due to its shared contents with the KNXP Group; following his analysis of common errors in
KNPX, Schwan even suggested that for a number of individual lyrics, V stood much nearer to the original than the other sources.228 In describing the various sections of V,
the traditional mode of inquiry was to establish the nature and ordering of its contents. In fact, the compilation of texts gives us little insight into the workings of the music notators, as it must have been established by the time of their work. However, it does give us the constraints within which the music scribes must have operated when searching for songs within their exemplars. Furthermore, it formed the bedrock of evidence for the earliest stemmas proposed for trouvère chansonniers.
1.1 Philological Work on Ordering and Schwan’s Stemma
The first investigation of this kind into chansonniers was launched by Gustav Gröber for the Occitan and Julius Brakelmann for the Old-French sources, both of whom
228See Schwan, Die altfranzösischen Liederhandschriften, pp. 112–113. Also, ‘Wenn wir die
Ergebnisse dieser Betrachtung zusammenfassen, so sehen wir, dass V dem Orginal viel näher steht, als gemeinsamen Vorlage teilt, sowohl in Bezug auf den Bestand wie auch in der Anordnung der Lieder.’
identified similarities between various groups of chansonniers.229 The most significant contribution here for our purposes is Brakelmann’s grouping of K, N, P
and X, whose shared ancestry remains unchallenged.230 Brakelmann’s view of V
derived primarily from his consideration of the religious contrafacts of V2, relevant to his particular interest in C (CH-Beb 389), the only textual concordance for many of
V2’s pieces.231 Brakelman’sinsight into the main body of the collection was therefore limited. Schwan’s 1886 summary of trouvère chansonniers was by far the most comprehensive and influential contribution of the 19th century; he established ordering as the second part of his three-pronged method for comparing trouvère manuscripts and introduced Lachmann’s ‘common-error method’ into trouvère source history.232 It was he who first attempted to assign V a place in the broader history of chansonnier production. His three avenues of inquiry corroborated each other: the contents of each volume, the ordering of the songs, and divergent textual readings. In all three of these areas, V shows an affinity for the so-called KNPX group. It is only when musical considerations are brought into play that V begins to be considered a peripheral source, as we shall see below.
Among other similarities, the members of the KNPX group share roughly the same sequence of songs. In his list, Schwan noted that in all 207 poems share their
229
Gustav Gröber, ‘Die Liedersammlungen der Troubadours’, Romanische Studien 2 (1877), pp. 337– 672; Jules Brakelmann, ‘Die dreiundzwanzig altfranzösischen Chansonniers in Bibliotheken
Frankreichs, Englands, Italiens und der Schweiz’, Archiv für das Studium der neueren Sprachen und Literaturen 42 (1868), pp. 43–72.
230
Brakelmann, ‘Altfranzösischen Chansonniers’, at pp. 51–4.
231Idem,p. 46: ‘Die 22 Lieder, die Lavallière 59 und B. 389 gemeinsam haben, zeigen nur geringe
Varianten, so dass eine gemeinsame Quelle beider Handschriften angenommen werden muss, wenn wir nicht gar annehmen wollen, dass dem Schreiber des Manuscripts Lavallière die berner Handschrift vorlag. Dieses, nicht das umgekehrte Verhältniss würde anzunehmen sein wegen des höheren Alters
ordering between V and the KNPX group.233 Using strict criteria for identifying agreement in ordering, counting inversion of any two songs and any interpolation of a song as a disagreement, I arrive at a slightly different number, 165 songs.234 In all, 162 songs in V follow the arrangement in K exactly, while only 112 follow the arrangement in N. This is mostly due to 43 songs which appear in the same sequence in V and K and are entirely missing from N. The discrepancy could be explained by one or more missing gatherings between fols. 8 and 9 of N, as proposed by Schwan and reaffirmed by Wallensköld in one of the earliest editions of Thibaut de Navarre’s works.235 The missing gatherings might easily have contained songs 17–26 and 28– 60.
Ultimately, ordering cannot determine whether V is more closely related to K
or to N, nor would it necessarily be revealing if it could. Hans Spanke, as we will shortly see, has thrown the possibility of establishing filiation within the KNPX group into serious doubt.236 The evidence is, in any case, equivocal: there are twelve instances where V follows K against N, and eleven instances where V follows N
against K. In addition, there are six additional songs in V that follow the same sequence as in N but are separated by the insertion of unica that appear only in V. Finally, there are an additional three pairs of songs that are swapped from their order 233 Idem, p. 113. 234 Songs 1–26, 28–57, 60–77, 78–9, 81–4, 85–98, 99–104, (only in N,99–102 in K), 105–12, 114–16, 159–63, 172–4, 175–7, 183–4, 185–97, 198–201, 202–11, 216–28, 229–31, 234–5, 261–2, 263–6 (only in N, K interpolates), 269–76 (only in N, K skips), are found with the same ordering in at least one of the KNPX sources. If interpolations in V are ignored and we count instances where three consecutive songs in V can be found in the same sequence in K or N but spread out by interpolations there, the number is higher. For the purpose of discussing ordering, songs will be referred to according to their number in my own catalogue for V in Appendix 1.
235Schwan, Liederhandschriften, pp. 96–99; Axel Wallensköld, Les Chansons de Thibaut de Champagne, Roi de Navarre (Paris: E. Champion, 1925), p. 97.
236Hans Spanke, Eine altfranzösische Liedersammlung: der anonyme Teil der Liederhandschriften
KNPX (Halle: Max Niemeyer, 1925). Neither Spanke nor subsequent scholars have challenged the fundamental similarities of these manuscripts or V’s relation to them in terms of contents, ordering, and textual similarities.
in K and N but still at least appear juxtaposed in V. Adding all of these songs to the 184 that follow K exactly, we arrive at Schwan’s figure of 207 songs that match the ordering of K and/or N. The number of all shared contents between V and K/N, regardless of ordering, is 228. Most of these additional 21 songs are witnessed by numerous sources outside the KNPX group. To address the question of whether V
reordered its exemplar or simply copied some songs from different sources, Schwan needed to turn to the third method, the comparisons of textual readings (Lesarten).
The consideration of textual variants adds considerable nuance to Schwan’s account of the relationships between KNPX and V, though his subsequent conclusions rely on dated assumptions. By providing examples of errors found in KNPX, but not present in V, he argued for a more direct link between V and the ultimate goal of his study, the Original.237 Schwan also assessed the insertions in K and N not found in V. These tend to be attributed to more recent authors, and Schwan suggested V’s exemplar was compiled before these poets were active, though he simultaneously dated V to later in the 13th century. In his stemma, therefore, Schwan placed the V
exemplar on a higher branch of the tree than KNPX, deriving from the putative source, φ. As the stemma depicts, this source in turn was copied to produce λ, nearest the common ancestor of KNPX.238
Figure 4.1: Schwan’s Stemma for the “sII Manuscript Family (re-created from p. 171 of Die altfranzösischen Liederhandschriften)
sII ρ φ σ β V λ O S R3 B L ν N x K ξ X P
As in many sources, the songs in V are roughly grouped according to author. In
KNPX, attributions are given in rubrics at the beginning of the first song of each author’s section. In addition to noticing the lack of attributions in V, Schwan also identified that V’s divergences from the ordering in KNPX cut across these shared groupings. In addition to swapping some author sections, V even removes whole sequences of pieces from one ‘author section’ only to place them after another.239 This disruption of the authorial groupings is neither jarring nor difficult to explain, due to the complete lack of attributing rubrics in V. Clearly, argued Schwan, the compiler of V was unaware of the groupings, or deliberately neglected them in the
239
process of assembling the volume. From this information, Schwan concluded that the compiler of V lacked interest in literary history and took a purely aesthetic interest in the songs.240 The re-ordering across author sections presumably occurred between V’s exemplar (φ) and V. K and N then made much smaller alterations to the ordering of φ, which consisted in moving certain songs to and from the anonymous section of unattributed pieces, an issue subsequently treated in more detail by Hans Spanke.241
1.2 From Spanke’s Der anonyme Teil to Gennrich’s ‘Repertoiretheorie’
Spanke’s edition of the anonymous part of KNPX (what he referred to as ‘K2N2P2X2’) did more than fill out the detail Schwan glossed over. His work shed doubt on Schwan’s entire method, without challenging the close relationship of the four sources per se. For Spanke, even before the question of oral transmission entered the picture, the rapid circulation of songs and their supposed independent proliferation on lost rolls prior to their collection in the ‘Liederbücher’ rendered the goal of a secure stemma unattainable.242 Referencing Bédier’s criticisms of Lachmann’s followers, Spanke offers concrete examples of independent errors contradicting Scwhan’s stemma, as well as citing the possibility of contamination (indeed, part of the difficulty is differentiating contamination from independent error).243 Spanke takes the familiar Lachmannian precept (that reasonable variants provide equivocal evidence and must be corrected by use of the stemma) and pushes it past its breaking point. Like Bédier, he points out the danger of multiple, variable redactions
240
Ibid. p. 246.
241
Spanke, Anonyme Teil, pp. 264–7, 269–71.
242
attributable to the author.244 Given what Spanke knew of contamination and what he believed about the likelihood that multiple authorial Pergamentblätter had once existed, any and all acceptable readings found in the extant sources could descend from an alternative originally invented by the author.245 Spanke therefore turns primarily to codicology and to the ordering of songs in the extant collections when establishing filiation and leaves the comparison of variants to one side. He considers the establishment of a stemma codicum that would be valid for each song, based on shared errors to be ‘untunlich’.246
In addition to these general methodological points, Spanke points out a number of specific errors on Schwan’s part. Some of these relate to the examination of variants, where Schwan had located errors in only two sources, when in fact they could be found in three (through mere oversight, or perhaps bias, the Bédierist argument is as relevant here as ever).247 Most important is Schwann’s ignorance of a crucial codicological point, a lacuna in K and N corresponding perfectly to a gathering break in each source. The songs Schwan had labeled c–u found only in X could therefore not be used to differentiate X from K and N. The piece that precedes song ‘c’ in X is found truncatedin both K and N, in slightly different places (verses 39 and 5, respectively).248 The truncation occurs at the break between Gatherings 19 and 20 in K and Gatherings 18 and 19 in N. Gathering 19 of N thenresumes with another song not found in K and finally returns to K’s ordering. Spanke notes also the similarities in decoration between K and N, as well as their otherwise similar ordering and the near synchronicity of their gathering structure (the second gathering of N is
244
Ibid. p. 276 and Bédier, ‘Classement’, at p. XXXVII.
245
Ibid. p. 274 and above for Spanke’s adherence to Gröber’s theory of Liederblätter.
246Ibid. p. 276. 247Ibid. p. 277. 248
missing so that its 18th gathering would presumably have been its 19th, as in K). From this he argues that both were prepared at the same place and time, and that the missing gathering in both must be attributed to an oversight in the delivery system. Due to this accident, two gatherings almost identical in terms of contents failed to reach the manuscript’s ‘buyer’ (Besteller).249 Spanke presumably believed the copied gatherings were mislaid prior to binding. It is this circumstance more than examination of variants that leads Spanke to unite K and N on his corrected stemma, though it cannot establish the placement of X in relation to them.250
Numerous difficulties in establishing filiation within the K2N2P2X2 group still remain for Spanke. Schwan had already noted that a number of songs found in K2
might be found with author attributions in other sources and offered the explanation (accepted by Spanke) that the scribe simply lacked the literary-historical interest to pursue the ascriptions. This is of particular relevance to V: a number of the songs found with author ascriptions in N1 and without in K2 (87, 90–92, and 140 by Spanke’s numbering) are also found in our source.251The existence of these songs without author attributions in K (and, above all their appearance in V) leads Spanke to posit a specific process of contamination, where (despite K’s primary filiation) K’s scribe consulted an exemplar shared with V but not with N2.252 This would be in keeping with later unica in K, which Spanke considers were assembled by the K
scribe in order to amplify the manuscript, and re-ordered more-or-less alphabetically to remedy the lack of author attributions.253
249
Ibid. p. 267. Spanke’s ideas of the process of commission and production are never made explicit; it is possible he was already aware of the co-ordinating activities of the libraire.
250
Ibid. p. 278.
Though V lacks these unica, it does contain four of the songs in K2 mixed in with its own unica, three of them in the same order as in K2 (Spanke’s 187 and 190– 192 = 260–263 in V). This complicates matters both for K2N2P2X2 and Spanke, and for our consideration of V. How much of the latter source actually shared an exemplar with K? And was K alone in consulting multiple earlier collections or might V have done the same? These questions form the crux, prompting the cross-comparison of codicology, ordering and melodic variants in V to be conducted below.
Spanke’s broader theoretical point amounts to the argument that the proliferation of fragmentary sources clouds the stemmatic exercise beyond all redemption. This view was appropriated and strengthened in the revised model of transmission proposed by Friedrich Gennrich and developed by Hendrik van der Werf. In the first part of the 20th century, the paradigm proposed by Gustav Gröber had dominated the discourse about the transmission of troubadour and trouvère song.254 Gröber divided transmission into two stages. The first period started from the troubadours’ own flourishing when, Gröber believed, they employed ink and parchment as part of the compositional process to stabilize (fixieren) their musical and poetic thoughts. This era was characterized by the circulation of songs individually in easily-transportable written form on sheets of parchment, called breus de pergamina
in one tornada and dubbed ‘Liederblätter’ by Gröber.255 The second period saw the collection of these sheets and rolls into larger collections, the Liederbücher some of which survive to the present day in the form of the chansonniers (Gröber follows his analysis with detailed descriptions and evaluations of each of the troubadour
254Gröber, ‘Die Liedersammlungen’, p. 342. 255
chansonniers). Both Schwan and Spanke had this model in mind when developing their stemmatic arguments regarding the trouvère collections.
By the time of Spanke’s Anonymer Teil, Friedrich Gennrich was already staking out an alternative position which was to supersede that of Gröber. The lack of discoveries of any Liederblätter in Old French had already been a topic of puzzlement for Schwann, who still asserted that such Liederblätter had nonetheless existed in ‘France’ (Frankreich).256 Gennrich, citing first of all this lack of evidence, denied the possibility that vernacular monophony could have been transmitted primarily in written form, arguing instead for extensive networks of oral transmission.257 Gröber was of course aware of the existence of joglars and jongleurs and acknowledged the co-existence of oral transmission, but assumed that the troubadours and trouvères (educated in writing as he believed they were) must have left records of their work. Gennrich most systematically refuted this position in a 1956 article where he proposed a Repertoiretheorie to rival the ‘sogen[annte] Liederblätter-Theorie’.
He attacked Gröber on several fronts. First, Gennrich argues against the literacy of the trouvères.258 The medieval biographies Gröber had relied on, which do indeed claim literary education for each of the troubadours in turn, are completely formulaic, each providing a variation on the theme of literary praise for their subject, for example saup ben lezer e chantar, saup ben trobar e cantar or amparet ben letras
for Piere Cardinal, Guillaume IX, and Arnaut Daniel, respectively.259 Secondly, he deploys evidence from the songs themselves where, thanks to the existence of envois
and tornadas, he can easily claim that the incidence of performers on hand and ready
256
Schwan, Liederhandschriften, p. 263.
to transmit songs through sound outstripped opportunities to inscribe music in parchment by far; indeed, according to Gennrich’s reading, many of the envois
demonstrate that even literate troubadours wrote their songs down only when no messenger happened to be available. Finally, Gennrich’s comparison of the extensive variants (‘weit gehende Abweichungen’) convinced him of the implausibility of autograph copies.260 In lieu of scattered leaves collected into books, Gennrich proposed that the first trouvère song ‘collections’ were the repertoires of individual singers, who learned pieces sometimes from the mouth of the first messenger to whom the composer had entrusted them, sometimes through longer strings of oral transmission. The peripatetic lifesyle of these jongleurs enabled the songs to travel long distances within a generation, or simply criss-cross frequently within a fairly