Transcribing Petrarch’s Genres in the Late Fourteenth Century

In spite of its apparent ‘betrayal’ of Petrarch’s visual poetics due to its one-column layout, the copy of the Rerum vulgarium fragmenta contained in MS Cologny Bodmer 131 documents a way of employing common graphic devices that still shows awareness of the graphological distinctions among the five genres of the Fragmenta, and yet unavoidably slips in occasional errors deriving from the loss of Petrarch’s original mise en page. Gambalunga Sc-Ms. 93, in its first unit, reproduces the same choices of organization of the genres, and often even the same errors, demonstrating its derivation from the same antegraph as the Bodmer manuscript. The two copies, both from the end of the fourteenth century, provide evidence that such transcriptional choices were already consolidated shortly after the death of the poet, suggesting that scribes of the earliest copies of the Rvf did not follow the formats of their holograph. “As we think about the possibility of literary history and about the meaning of textual authority in relation to Petrarch, it will be increasingly necessary to take into account as many of the scribal versions of his poems as we can” (Del Puppo 2004, 131). With these words, Dario Del Puppo was sending an open invitation to further proceed on the inspection of the “evolving transcriptional pragmatics” of Petrarch’s poems in the fifteenth century (2004, 130). MSS Bodmer 131 from the Fondation Martin Bodmer of Cologny (Geneva) and Sc-Ms. 93 from the Biblioteca civica Gambalunga in Rimini are two copies of Petrarch’s Rerum vulgarium fragmenta produced in the Veneto in the late fourteenth century that allow us not only to broaden the spectrum of “witnesses of how he [Petrarch] was read and understood” (Del Puppo 2004, 130), but also to redate to the last Textual Cultures 13.1 (2020): 106–127. DOI: 10.14434/textual.v13i1.30074 G. Benghi : Transcribing Petrarch’s Genres in the Late Fourteenth Century | 107 decade of the Trecento the time of that widespread circulation of Petrarch’s lyric poetry when scribes undertook a variety of editorial choices, altering the poet’s unique “visual poetics” within just a couple of decades from his death.1 The two manuscripts have been dated from the end of the fourteenth century in their respective catalogues, and, in agreement with the given dates, I intend to demonstrate the importance of dating these two codices by the end of the fourteenth century, especially for their copyists’ editorial choices in altering Petrarch’s poetics.2 The anonymous scribes of these two copies, in fact, align the text in the single column, one-verse per line format that “took root at the end of the Trecento”, a transcriptional style that became the norm of “the vulgata in the Quattrocento” (Del Puppo 2004, 128). The popularity of this, so called ‘modern’, layout in the fifteenth century is renowned, while it might have been underestimated by literary historians for the fourteenth century. In the codicological examination made by Marco Cursi (2014) of the earliest manuscripts of the Canzoniere (Rerum vulgarium fragmenta, or the Rvf) we notice that the group of codices with layout “ad una colonna” is made up of only six manuscripts out of the total of 29 dated as “antiquiores”, and that only three out of those six are datable to the end of the fourteenth century.3 However, our MSS Bodmer 131 and Sc-Ms. 93 can be rightly added to the group, as they fully respect the two criteria applied by Cursi to his examination — the two criteria being 1) a dating within the first decade of the fifteenth century, and 2) the nature of the collections as books, and not as miscellanies.4 By adding Bodmer 1. The term “visual poetics”, coined in the 1980s by Storey (1989; see also 1993) and now in wide usage, refers to the use of transcriptional and typeset layouts in manuscripts and printed works as part of the intentional poetics of a literary composition. 2. For MS Bodmer 131, a detailed codicological description by Allegretti 2003 and the digital version of the codex are available on the website of the “Virtual Manuscript Library of Switzerland”: https://www.e-codices.unifr.ch/it/description/fmb/cb-0131/. For Sc–Ms. 93, instead, I refer to the two descriptions available on the website of the “Biblioteca Digitale Italiana”, by selecting “Biblioteca Civica Gambalunga” and searching the codex with its previous shelf-mark, that is D. II. 21, in both “Catalogo Lucchesi” and “Catalogo Nardi”: http://cataloghistorici.bdi.sbn.it/indice_cataloghi.php. See also the Images in my Appendix for SC-MS 93, pp. 123–24. 3. See the “Tabella 5 – Codici dei Rvf ad una colonna” in Cursi 2014, 244. 4. Cursi (2014) suggests that “ammesse nel corpus le raccolte palesemente orientate verso l’adesione ad un modello di libro lirico d’autore, pure se aperte a contributi 108 | Textual Cultures 13.1 (2020) 131 and Sc-Ms. 93 to this group of codices of the Fragmenta as classified by Cursi, the number of manuscripts in one-column layout and dated to the end of the Trecento increases from three to five. When we compare the other charts made by Cursi with all the types of layouts employed by the codices antiquiores, but confining the analysis to the last quarter of the fourteenth century, we realize that the number of codices with one-column layout (5) is almost equal to the number of codices with other types of layouts: the manuscripts with a two-column layout but with the verses copied in vertical order rather than horizontal (“a impaginazione verticale”, Cursi 2014, 242) are also 5 (6 when including MS London British Library, King’s 321 dated “anno 1400” by its Venetian copyist), while the number of codices employing a two-column layout but with a variety of copying solutions between horizontal and vertical readings (“a impaginazione mista”, Cursi 2014, 240) are 3, or 4 if including MS Firenze Biblioteca Laurenziana Segniano 1, dated to between the end of the fourteenth and the early fifteenth century first by Storey (2006, 295) and later again by Cursi (2014, 240). The predominant group of manuscripts remains those that transcribed Petrarch’s lyrics in what Cursi would later call an “impaginazione nobile” (i.e., in two columns, reproducing Petrarch’s mise en page [2014, 238–46]), which are six — eight when including MS Firenze Biblioteca Laurenziana Pluteo 41.10 and MS Paris Bibliothèque nationale de France, Italien 551, actually dated by Cursi between the fourteenth and fifteenth centuries.5 Thus, while the use of the two-column layout shows to be prevalent overall in this early period, when we focus on the variety of changes to Petrarch’s ‘noble’ mise en page already employed by the scribes at the end of the fourteenth century, we realize that the one-column layout was not simply the outcome of a progressive corruption of Petrarch’s visual poetics, but rather part of a spectrum of editorial choices available to the scribes, behind whose choices there were specific reasons caused by the circumstances of the copying process and/or imposed by the terms of the manuscript’s commission: a situation well described by Storey with reference to the first half of the Quattrocento,6 but that may be already happening by extravaganti, escludendo invece i testimoni riconducibili alla tipologia del canzoniere miscellaneo”, and limits the manuscript’s dates “dagli anni in cui l’autore era ancora vivente fino al primo decennio del secolo XV” (229). 5. For a prolonged description of the Petrarch’s two-column layout and its philological and interpretative implications in his visual poetics, see Storey 1993, 201–419. 6. “una diffusione del Canzoniere multiforme già nella prima metà del Quattrocento, ormai inestricabilmente legata alla variabile di copia e alle scelte ‘ediG. Benghi : Transcribing Petrarch’s Genres in the Late Fourteenth Century | 109 the last decade of the Trecento, for we see in the comparison above that some of the eminent copies of the Rvf, such as MS Laurenziano Pluteo 41.10 and Paris Bibliothèque nationale de France, Italien 551, date from slightly later than other manuscripts with the one-column layout like the Bodmerian and Riminese codices.7 The evidence offered by witnesses such as MSS Bodmer 131 and Sc-Ms. 93 can therefore address this research along the paths indicated by Brugnolo, Storey, Del Puppo, and other scholars,8 which is to inspect the formats of the early copies of the Fragmenta in order to verify what they can tell us about their antegraphs and the earliest circulation of Petrarch’s poems. As Del Puppo points out, “Wilkins does not discuss whether the poems in earlier forms of the Canzoniere are laid out as they are in the holograph, Vaticano latino 3195, and whether scribes would have followed these formats” (2004, 116), while this is, in fact, a crucial point that awaits further investigation. Before proceeding, I must outline the unique list of features that Bodmer 131 (for which I will use the initial “B”) and Sc-Ms. 93 (SC) have in common. Codicologically, they depict the favorite forma-libro, or book form, of the earliest tradition of the Canzoniere, which is an average-sized book with gothic or semi-gothic script on parchment.9 By coincidence, they are both composed of two different units: in B, the first section (cc. 8r-143v) contains the Rerum vulgarium fragmenta plus 36 mostly interspersed Distoriali’ dei vari copisti, siano esse frutto di adattamenti (o fraintendimenti) di comodo del singolo amanuense o il riflesso di una nuova esigenza socio-culturale dettata dalla committenza”, in Storey 2006, 300. 7. While Del Puppo describes MS Laurenziano Pluteo 41.10 as “made during the last two decades of the fourteenth century in the Veneto” (2004, 121), Cursi suggests that the manuscript was produced sometime between 1397 and the first years of the Quattrocento (Pulsoni — Cursi 2010, 258–68); Teresa De Robertis recently dated it “circa 1400” (2016, 67). 8. See especially Brugnolo 1991 and 2004, Pacioni 2004, and Del Puppo 2004 and 2007. 9. Cursi generally describes the standard format for lyric transcriptions: “per le Rime il modello librario preferito prevedeva il materiale scrittorio membranaceo, misure medie, scritture gotiche o semigotiche” (2014, 233). MS Bodmer 131 measures 230 x 163 mm, with a writing space of 32–33 lines, while Sc-Ms. 93 is 210 x 113 mm, with a ruled writing space varying between 24 and 26 lines in the first part and 23 in the second part. As for the script, it is a littera textualis; Pancheri (1993, 78) claims that it has a touch of bastarde. 110 | Textual Cultures 13.1 (2020) perse (or Estravaganti),10 while the second section (cc. 144r-178r) contains 29 poems by Dante plus seven by Guinizzelli, Cavalcanti and Cino; in spite of its position at the end of the book, the script and the construction of the quires suggest that the second section was transcribed earlier than the first one.11 On the other hand, what undermines the unity of MS SC is not the content but an abrupt change in the script and in the mise en page, on c. 61v (in the midst of Rvf 181, v. 4, see my Appendix Image 1 and Image 2), due to the intervention of a second scribe, so that the manuscript must be considered as two units, independent from one another both in their codicological and textual aspects; these two units I will indicate as “SC(a)” and “SC(b)”. A significant common trait between B and SC, instead, is the identical sequence of the poems — though only up to Rvf 215, and excluding the mechanical lacunae both at the beginning, at the end, and in the midst of the collection.12 In addition to containing the same four10. “Disperse” (uncollected) and “estravaganti” are the terms usually used for those poems not included by Petrarch in his holograph (MS Vatican Latino 3195) but attributed to Petrarch by the manuscript tradition. The first and only complete edition of all these spurious poems was edited by Angelo Solerti in 1909, and recently republished in fac-simile with an introduction by Vittore Branca and an Afterword by Paola Vecchi Galli (see Solerti 1909 and Vecchi Galli 1997). 11. I provide a more detailed analysis of the Dante and Stilnovo section of MS Bodmer 131 (B) in Benghi 2020. For the convenience of online and remote readers, references to the Bodmer manuscript will be to the most immediate access to the codex in its digital form in the e-codices site (see n.2 above). This means that in some cases the original numbering of the chartae in the upper right-hand corner of the recto of each charta will not be reflected in the digital access information for the manuscript due in part to a missing charta between the index (cc. 1r–7v in the original numbering) and its initial leaf (c. 9r in its original numbering = digital c. 8r [note the reference on c. 7v in the ancient index to Voi che ascoltati in rime sparse il sono [sic] as c. 9]). Those who consult the codex directly in Cologny should adjust the numbering accordingly. 12. See the Appendix below with the complete sequence of the Rvf in both manuscripts. Please note that the numbers assigned correspond to the order revised by Petrarch by means of marginal numbers in his holograph MS Vatican Latino 3195. Thus my numbers are different from those of the physical order in which the last 31 poems appear in the manuscript, the order that was adopted by Wilkins (1951) and in the Petrarchive (Magni, Storey and Walsh, 2014– 2020), the latter of which indicates Petrarch’s revised order as “rev”. Thus the final sequence of B, described by Wilkins (1951, 232) for what he knew as “Melziano A” (the codex resided in the Melzi Library in Milano at the beginning of the twentieth century), was: “356–360, Un clima, c, cxxix, 361, 366” is in its G. Benghi : Transcribing Petrarch’s Genres in the Late Fourteenth Century | 111 teen Disperse copied in the same position, the two manuscripts share — in fact — the same inversions in the two sequences Rvf 90, 89, 91 and Rvf 153, 155–158, 154.13 Finally, a comparison of the lectiones reveals that B and SC share numerous variants (varianti congiuntive), while also containing a minority of different readings (varianti disgiuntive); therefore, we deduce that they must descend, even if not directly, from the same antegraph (as documented also by their sequence), while we must exclude that one is copy of the other. An examination of their mise en page confirms this close relationship between the two manuscripts, which, based on an analysis of the variants, I have identified as similar to two ‘cousins’: they share the same system of graphic features, and sometimes even the same mistakes, to reproduce the metrical schemes of the Fragmenta, proving that their formats derive mostly from their antegraph, and yet the two copies also contain independent mistakes by either one of the scribes.14 All this is true, however, only pertaining to the first section of SC, that includes Rvf [*44]–181 (Rvf 1–43 are missing due to mechanical loss of the chartae), since the change of scribe on c. 61v coincides with a change of the antegraph — as a change in the lectiones also confirms.15 MS Bodmer 131 is a very clean, finished product, which displays lavish illuminations of leafage and dragons on its opening charta (c. 8r), and has each poem initial in red ink with blue entrelacs, while all capital letters are marked in red. The Riminese codex, in its first section, is instead completely devoid of all rubrics and decoration, leaving empty spaces for the missing decorated initials (of the size of three or four writing lines, as in B). In spite of the different aesthetics, the scribes of both B and SC(a) organize all five genres of the Canzoniere in the same one-column, one verse per line format, and distinguish the five poetic genres of the Fragmenta through the use of capital letters, that are always located slightly toward the left margin (and that, in B only, are marked in red), in order to separate the metrical units of each genre. It is useful for the modern reader to remember that these capital letters do not follow a period, and therefore do not express the real syntactic flow of the discourse, but serve purely as metrical markers corrected and revised order (reported in my Appendix): 360–364, Disp. 50, Disp. C, Disp. CXXIX, 365–366. 13. I refer to the Disperse by the number in the order assigned by Solerti 1909. These are the fourteen Disperse contained both in B and SC-MS 93: CCXIII, XXV, CC, CLXXXII, XXVII, XXIX, XIX, LXIV, CX, XLVIII, L, C, CXXIX. 14. The nomenclature suggests that they are copies of two different exemplars that, in turn, derive from the same antegraph. 15. For an extended examination of this topic, see Benghi 2020. 112 | Textual Cultures 13.1 (2020) (in any case, in B and SC, punctuation is rare, and mostly represented by the punctus). Among the five genres, the sonnet stands out for its typical, unchanging structure of 4+4+3+3 verses, resulting from the identification of the quatrains and the tercets by the capital letter that introduces the first verse of each metrical unit. As modern readers, we tend to take this layout for granted, failing to realize that in this way the constituent unity and visual solidity on which Petrarch’s sonnet was conceived is completely disrupted,16 for “Petrarch’s sonnet format is to assure integration of the tercets into the body of the sonnet by not spatially distinguishing them from the octave” (Storey 1993, 238). A brief digression needs to be opened here to recall that the breakage of the sonnet’s unity as designed by Petrarch is already found in some of the ‘faithful’ copies such as MS Laurenziano Pluteo 41.17, where a paragraph mark is placed next to the first tercet, or in MS Laurenziano Segniano 1, where the capital letter of v. 9 (that is the beginning of the first tercet) is taller than all the other versali, occupying two writing lines. Not only the separation of the tercets from the octave was common practice, deriving directly from the poetic tradition of the thirteen century,17 and applied independently from the type of layout (for instance, in MS Trivulziano 1091, that has a mixed layout,18 the capital letter of v. 9 is marked in red19), but also the breakdown of the tercets seemed somehow 16. “La forma progettata del cod. Vat. lat. 3195 è infatti concepita per essere visivamente e semanticamente organica, unitaria e compatta nel suo svolgimento in progressione dall’ottava (le due quartine) alla prima terzina e soprattutto da questa alla seconda terzina”, in Storey 2004, 153. 17. As Storey explains, “l’impostazione petrarchesca riprende modalità di mise en page codificate dal duecentesco cod. Vat. Lat. 3793, tralasciandone i marcatori paratestuali più superficiali, quali le graffe per le distinctiones più difficili” (2004, 153). As a matter of fact, the scribes of MS Vatican Latino 3793 tend to place a paragraph mark next to the first tercet, and also emphasize the overall structure of the sonnet as composed of an octave plus two tercets by tracing two brackets on the right side of the sonnets, with the first bracket enclosing the two quatrains (vv. 1–8) and with the second bracket enclosing the two tercets (vv. 9–14). 18. “ad impaginazione mista” (Cursi 2014, 240). This codex can be seen on line, thanks to the “Manus Online” project: https://manus.iccu.sbn.it//opac_SchedaScheda.php?ID=50126. It contains Petrarch’s Fragmenta and also the fifteen canzoni distese, as we find in MS Bodmer 131. But in MS Trivulziano 1091 Dante’s songs are at the beginning of the collection, before the Rvf. 19. Starting from c. 10v, in Rvf 12. In this sonnet, we actually see a red paragraph mark next to v. 9, presumably to point out even more clearly the beginning of G. Benghi : Transcribing Petrarch’s Genres in the Late Fourteenth Century | 113 necessary to the eye of most readers — as Boccaccio’s renowned copy of the early version of the Fragmenta, the Fragmentorum liber, in MS Vatican Chigiano L.v.176 witnesses, with the regular red and blue paragraph marks that introduce respectively the first and the second tercet of each sonnet. As we know, Boccaccio’s copy dates from at least twenty-five years earlier than the other ancient manuscripts so far taken into account (and also embodies the gothic culture Petrarch strived to reform with his revolutionary visual poetics), therefore it might seem an inappropriate example.20 However, when we look again at MS Trivulziano 1091, dated 1391–1410 by the catalogue of the Trivulziana Library, we find that the copyist felt the need to separate the second tercet from the first one by placing two oblique dashes between vv. 11 and 12: in other words, “il copista applica una ‘traduzione culturale’ per cui impiega una rete — a volte anche microscopica — di meccanismi per rendere la veste grafica del sonetto il più leggibile possibile per un pubblico attuale” (Storey 2017, 97). These comparisons help us realize even better how Petrarch’s visual poetics, in his holograph MS Vatican Latino 3195, was truly unique, and how “la ripresa da parte di Petrarca dell’antico stile dei Siciliani [. . .] risulta già una formula considerata arcaica nel primo Quattrocento se non prima” (Storey 2017, 97 [my italics]). Going back to the graphological organization of the genres in B and SC(a), ballate and madrigali are also identified in their metrical schemes by capital letters, or initials, so that they are distinguished from the sonnets by their variable units (and length), reproducing rather faithfully Petrarch’s metrical experimentation. In B, for example, it is striking to find that the copyist doesn’t fall into the error of confusing the two ballate grandi (Rvf 11 [Lassare il velo o per sole o per ombra] and Rvf 14 [Occhi miei lassi, mentre ch’io vi giro]) as sonnets, in spite of the same number of total verses (14): in these two poems the capital letters are placed only in vv. 5 and 11, resulting in the scheme 4+6+4, that resembles the structure of the ballata grande (refrain [4 verses]+piede [3 verses]+piede [3 verses] +volta [4 verses]), with the only exception of not dividing the two piedi. Nevertheless, with the ballate we begin to see how a less sophisticated reproduction of Petrarch’s poetics inexorably gives way to confusion. In ballata mezzana Rvf 55 (Quel foco ch’i’ pensai che fosse spento, c. 30r), the capital letters are placed at the beginning of vv. 4 and 11, establishing a structure of 3+7+7 verses that reveals that the copyist of B identifies the refrain from the body of the ballata, while the tercet, but this practice remains an exception. 20. See the extensive work by Storey both on Boccaccio’s conservative practice of copy and on the uniqueness of Petrarch’s visual poetics in Storey 1993 (and also in Storey 2015 for Boccaccio’s copy of the Fragmenta). 114 | Textual Cultures 13.1 (2020) he does not pay attention to the separation of two piedi from the volta. In Rvf 59 (Perché quel che mi trasse ad amar prima, c. 31r), another ballata mezzana with the same scheme of Rvf 55, the copyist instead begins v. 4 with capital letter, and then again v. 6, 11 and 15, thus establishing the scheme 3+2+5+4+3, which blurs the original formula of the two piedi (2 verses+2 verses) followed by the volta (3 verses), repeated twice.21 Unfortunately, due to the loss of several chartae in SC, we cannot complete a comparison of these ballate. However, SC(a) does contain the two remaining ballate grandi, Rvf 63 and 149 (Volgendo gli occhi al mio novo colore, cc. 51r–v, and Di tempo in tempo mi si fa men dura, c. 70v), that share with B the same (correct) scheme, though in B we find a red paragraph mark next to v. 5 (i.e. the beginning of the piede after the refrain) of Rvf 149, that does not appear in SC(a).22 The case is particular, since neither in B nor in SC(a) did the scribes ever apply the paragraph marker to the genre of the ballata, while we remember that Petrarch uses it in the three ballate Rvf 55, 59 and 63, but not for Rvf 149.23 Moving to the four madrigals of the Rvf, the division of the verses substantially resembles the original metrical schemes: The only exception occurs in B’s transcription of Rvf 54 in which the copyist fails to distinguish the two couplets.24 The division units, indicated in brackets above, are as follows: 3+3+2 for Rvf 52 (c. 28r in B), 3+3+4 for Rvf 21. The metrical scheme of this ballata mezzana is as follows: XyY Ab Ab ByY Cd Cd DyY. 22. The breakdown of the verses in Rvf 63 is 4+3+3+4, which represents the metrical scheme XYYX ABC BAC CDDZ, and in Rvf 149 is 4+4+4+4, representing X(x)YyX AbbC AbbC CDdZ. 23. We should note, however, that Rvf 149 is the only ballata that employs mid-verse rhymes, rima al mezzo, in the refrain, an element that might play a role in this graphic solution by the copyist of B. We should also keep in mind that these poems were actually copied by Malpaghini (see Storey 1993, 264 and 2004, 158), whose work Petrarch oversaw and corrected. The layout of these poems in MS Vatican Latino 3195 can be consulted on the Petrarchive (see Magni, Storey, and Walsh: http://dcl.slis.indiana.edu/petrarchive/content/c013r-c013v. xml). 24. Rvf 52 and 54 are missing in SC. Rvf 52: Non al suo amante più Diana piacque: ABA BCB CC [3+3+2] Rvf 54: Perché al viso d’Amor portava insegna: ABA CBC DE DE [3+3+4] Rvf 106: Nova angeletta sovra l’ale accorta: ABC ABC DD [3+3+2] Rvf 121: Or vedi, Amor, che giovenetta donna: ABB ACC CDD [3+3+3] G. Benghi : Transcribing Petrarch’s Genres in the Late Fourteenth Century | 115 54 (c. 29v in B), 3+3+2 for Rvf 106 (c. 48r in B; c. 23v in SC), and 3+3+3 for Rvf 121 (c. 56r in B; cc. 34r–v in SC). When we compare the organization of the madrigals in B and SC with Petrarch’s mise en page in his partial holograph (Vatican Latino 3195), we can see how the later manuscript Figure 1. Vatican Library, Latino 3195, c. 23r; courtesy of the Biblioteca Apostolica Vaticana 116 | Textual Cultures 13.1 (2020) tradition exemplified by B and SC was based exclusively on the metrical schemes. Yet in the case of Rvf 106, c. 23r of Latino 3195, we note that Petrarch’s scribe Giovanni Malpaghini has copied the madrigal two verses per line (2+2+2+2) in order to stress the couplets instead of the two initial tercets (vv. 1–3 and 4–6), now distributed over three lines of transcription. While Petrarch’s own mise en page for Rvf 52 (c. 11v), Rvf 54 (c. 12v) and Rvf 121 (c. 26r) reinforces the prosodic and syntactic phrasing in these madrigals’ tercet structure, the layout of Rvf 106 emphasizes the meter and closure of the interlocking rhymes –orta (vv. 1, 4 [accorta – scorta]) and –ino (vv. 3, 6 [destino – camino]) that link internally the syntactic construction of the first six verses against the closing couplet (vv. 7–8 [poi – suoi]).25 In the case of the Bodmer copy, the scribe’s transcriptional system ignores the integrative value of Petrarch’s approved layout designed to underscore the madrigal’s variation. The 3+5 visual structuring of the madrigal in B instead reinforces the madrigal’s opening tercet, and, in opposition to Petrarch’s model, links the second tercet to the final couplet (see Fig. 2). The genre of the canzone in the Bodmer codex demonstrates instead two-line decorated initials that introduce every stanza, while the internal structure of the stanzas is conveyed by regular-sized and rubricated small initials. In essence, these decorated initials opening the stanzas actually distinguish the canzoni from the other genres, as opposed to the red paragraph marks employed for the stanzas of the sestina. This distinction is only true in B, since in SC(a) the sestina is simply organized by small initials at the beginning of each stanza. It is remarkable that in both copies the sestine, calculated by Petrarch as part of the canzone genre, are transcribed without errors, with their six stanzas succeeding one another in the correct order.26 We find one exception in SC(b) (cc. 120r–121r): the verses of Rvf 332 (Mia benigna fortuna) — the only double sestina (sestina doppia) of the Fragmenta — are copied in a strangely alternating sequence, that is 1, 24, 2, 25, 3, 26, 4, 27, and so on. This is a significant error for which a clear understanding of B’s and SC’s common ancestor can guide us. What has happened here is that the scribe of SC(b) has copied from an antegraph 25. Marco Pacioni (2004) remarks that even in Petrarch’s short madrigals variation in rhyme constructions are in invariably reflected in a nuanced mise en page (“la variatio rimica è così strutturante che è necessario renderla sempre evidente sul piano grafico, nella mise en page” [373–4]). 26. One error is found in B’s copy of Rvf 239 (c. 93v), in which the scribe fails to copy the congedo, which is then added by the same hand on the left margin of the poem, thus this case is to be considered simply a distraction of the copyist rather than due to the complexity of Petrarch’s original layout. G. Benghi : Transcribing Petrarch’s Genres in the Late Fourteenth Century | 117 whose transcription of the double sestina is in the same two-column mise en page as that of Petrarch’s authorized copy, MS Vatican Latino 3195, which lays out the sestina, uniquely among the five genres, in a vertical reading down the left-hand column and then continues the song’s remaining verses down the right-hand column, invariably producing two columns of unequal Figure 2. Cologny, Fondation Martin Bodmer, Cod. Bodmer 131, c. 48r (49r original number) 118 | Textual Cultures 13.1 (2020) length with a blank space at the bottom of the right-hand column.27 This change in Petrarch’s usual horizontal layout, from the left to the right column for all genres except the sestina, clearly caused some confusion with the SC copyist, just as it did with the scribe of Laurenziano Segniano 1.28 The copyist of SC read his exemplar from the left to the right column, just as he had clearly done in the case of the other poetic genres, converting their horizontal reading strategy into a vertical presentation. We should note one additional detail in SC: red paragraph markers are placed next to vv. 25, 7 and 31, which are the first verses of, respectively, the fifth, second, and sixth stanzas. When the copyist returns to rubricate his copy — for the rubricator’s hand coincides with the scribe’s — he identifies the correct beginnings of every stanza in spite of the wrong order of the verses. The use of paragraph marks in the sestina Rvf 332 is even more puzzling when we realize that this is the only case in which the rubricator applies them to a sestina (rubricated as “sonetto”), and that the only two additional places where the same red markings are found are not in sestine but in the canzoni S’i’l dissi mai (Rvf 206 [cc. 69v–70v]), rubricated as a “sonetto”, and Ben mi credea (Rvf 207 [cc. 70v–72v]) — rubricated as “cançona morale”. MS Bodmer 131 instead reveals an example of confusion between the genres of canzone and sestina, a problem originating perhaps with the disruption of Petrarch’s mise en page: on cc. 18v–19v, Rvf 29, Verdi panni, a canzone, which is punctuated with red paragraph markers reserved for the sestina next to each stanza, just like the poem that follows it, Rvf 30, Giovane donna, which is a sestina. But the unique prosody of Rvf 29, which consists of coblas unissonans, requires the same rhymes for each stanza, a repetition that might have caught the eye of the rubricator, leading him to mistake the canzone for a sestina (though the copyist doesn’t seem to mind that each stanza of the mistaken sestina has seven verses rather than the sestina’s required six). As we recall, the “struttura metrico-strofica” (Storey 2004, 162) of the canzoni is conveyed as well by the rubricated initials, by which the metrical sub-units of each stanza are identified, and, overall, faithfully represented. And yet, in addition to the loss of Petrarch’s subtle experimentation in 27. For a description of this standardized use in Petrarch’s visual poetics, defined by Storey as “descriptive sestina space”, see “Sestina” and “Spaces” in the Glossary of Magni, Storey and Walsh 2014–2020, http://dcl.slis.indiana.edu/petrarchive/content/glossary.xml. 28. The copyist of MS Laurenziano Segniano 1, from the first part of the Quattrocento, standardizes all the layouts of Petrarch’s lyrics in the Rvf to conform to Petrarch’s horizontal reading strategy, also applied to the sestinas. G. Benghi : Transcribing Petrarch’s Genres in the Late Fourteenth Century | 119 variation (variatio) based on similar metrical structures, copyists required to intervene between the two systems of the copy and his modern reader invariably misunderstood certain forms. In Rvf 127 (In quella parte dove Amor mi sprona), for instance, the copyists of both B (c. 59v–61v) and SC(a) (cc. 38v–40v) structure the first three stanzas with initials to create four internal units of 3+3+5+3 verses for the rhyme scheme: ABC BAC CDE eDe FF.29 But from the fourth stanza on the subdivision changes to 3+3+3+3+2, and then in the fifth stanza it changes again into 3+3+3+2+3 perhaps to reflect the changing syntactic structures of the stanzas. In the famous canzone Italia mia, benché ’l parlar sia indarno (Rvf 128, cc. 61v– 63v in B, cc. 40v–43r in SC), we notice even greater variation in how the copyists organize the metrical structure of each stanza whose rhyme scheme is AbC BaC cDE eDdf GfG. The first stanza is divided into units of 3+3+3+4+3 verses, the second in 3+3+3+2+2+3, the third like the first, the fourth is instead changed to 3+3+2+6+2, the fifth changes again to 3+3+3+2+3+2, and the sixth to 3+3+3+3+2+2 and 3+3+3+5+2 for the seventh and final stanza. It is noteworthy is that these variations are identical in both B and SC(a), as it happens for many other poems as well, such as Rvf 71, Perché la vita è breve, for which the copyists change the repeated 6+3+4+2 visual-syntactic division of the verses to 5+3+5+2 for the last stanza, or Rvf 72, Gentil mia donna, i’ veggio, where they change the same structuring division (6+3+4+2) into 3+3+3+6 for the last stanza. In Rvf 129, Di pensier in pensier, di monte in monte, along with the same changes in the subdivision of the stanzas, the two copies also share an error in the transcription, introducing the third stanza with a larger initial, as if it were the incipit of a new song. In fact, the copyist of SC(a) even leaves a line of blank space before this stanza. Yet only in SC(a) the fourth stanza is also introduced by a taller, three-line initial (instead of two), thus causing the stanzas of Di pensier in pensier (Rvf 129) to be split into individual poems (compare c. 64r in B with cc. 43v-44r in SC(a), see Appendix, Image 3). This and other occurrences of the same errors (congiuntivi) in the transcription (often sustained by the same variants in the text) are, at the same time, accompanied by independent errors (disgiuntivi), leading us to exclude the possibility that one manuscript may be a copy of the other. Rvf 126 offers, in this regard, one more excellent example: it contains one variation of the strophic division in common, and one instead unique to SC(a); moreover, the copyist of B omits v. 56, and adds it in the margin (c. 29. See Capovilla 1998 for a description of the metrical structure of Petrarch’s songs and for their rhyme schemes. 120 | Textual Cultures 13.1 (2020) 59v), while in SC(a) it is transcribed regularly in the body of the text (c. 38r); but in both copies the verse reads “orsi charco doblio”, differing from Petrarch’s holograph (“cosi carco doblio”). The second section of SC (to which I refer as SC(b)), nonetheless, must derive from another antegraph, proved by the readings and its very different modality of transcription: all the poems, still in a single-column layout, are copied with the head verse (versale) opening each verse (always traced in red), without deference to the prosody, and therefore with no distinction between genres. Moreover, all the poems are separated only by a oneline rubric (in the same hand) that reads “Sonetto di mess(er) franc(esc)o petrarcha” for every genre, whether it is actually a sonnet, short poem (as madrigals and ballate were sometimes deemed) or not, as it happens for the sestina Rvf 214 (on c. 74v), for all the ballate and madrigals, and for several canzoni, while there are only a few canzoni that carry the rubric “Cançona morale”.30 In the homogeneous and compact layout of SC(b), visually the opposite of the division of the metrical units of SC(a) and B, the few graphological devices employed to organize the prosody stand out, along with the inconsistency with which they are used. Rvf 206 and 207 (S’i’ ’l dissi mai, ch’i’ vegna in odio a quella and Ben mi credea passar mio tempo omai, cc. 69v–72v, the first two canzoni in our section SC(b)) are transcribed in a prose format, and with paragraph markers inserted only to signal the stanzas. By the same token, the same copyist transcribes Rvf 264, 268 and 270 (I’ vo pensando, et nel penser m’assale, Che debb’io far? che mi consigli, Amore? and Amor, se vuo’ ch’i’ torni al giogo anticho, c. 90r to 97v) rigorously one verse per transcriptional line, with the stanzas distinguished by the empty space for what we suppose to be decorated initials. Even more variations occur in subsequent canzoni: in Rvf 323 (Standomi un giorno solo a la fenestra, cc. 113r–114v) there is no separation of the stanzas whatsoever; in Rvf 325 30. We should keep in mind that it is common in many copies of the end of the fourteenth century to find a discrepancy between the differentiation of the genres by layout and graphic devices and their actual labeling: Petrarch’s poems are usually referred to, in fact, as either “sonetto” or “canzone” (usually accompanied by “morale”), as we read in the rubric opening the Rvf in MS Trivulziano 1091: “Sonetti echançoni morali di mess(er) francescho P(etrarca) poeta fiorentino” (c. 9v). It is interesting to find an individual label of the genre of the madrigale (expressed by the letter “m.” in opposition to “s.” and “c.” of all the other poems) in the index of MS Trivulziano 1015 (dated last quarter of the fourteenth by Cursi 2014, and accessible online through the Manus database: https://manus.iccu.sbn.it/opac_SchedaScheda.php?ID=229873). G. Benghi : Transcribing Petrarch’s Genres in the Late Fourteenth Century | 121 (Tacer non posso, et temo non adopre, cc. 115r–117r) the scribe leaves one empty line between the stanzas, and yet he has copied Rvf 325 immediately after the ballata Rvf 324, Amor, quando fioria, without an introductory rubric nor any space between the two poems. What determines these changes in copying the same genre? Was the scribe of SC(b) simply negligent, or was he perhaps sensitive to subtle variation? Can we conjecture that he was copying from a single antegraph for Rvf 206–207, from a different exemplar for the group Rvf 264, 268 and 270, and then from a third (and even forth) for the next songs? It is not very likely that a single copyist would have consulted three or four different exemplars to reconstruct a text like the Rvf.31 Rather it is more likely that some canzoni were read in a different way than others, and therefore reproduced differently. In this light, the reason why only a few songs were introduced by “Cançona morale”, instead of by “Sonetto” as the majority, could find explanation, though the — mostly wrong — identification of the genre of the canzone still leaves us with the question on whether it was taken from the antegraph or was rather generated autonomously by the scribe. MSS Bodmer 131 and Sc-Ms. 93 show, on one hand, how Petrarch’s antegraph/s must have played a key role in influencing the editorial choices of a specific format and graphological devices, while, on the other hand, these later copies’ usual, one-column format and graphic-metrical system was by the last decade of the fourteenth century already being tested and solidified. Comparing all the “forme strofico-grafiche” (Storey 2004, 166) of the canzoni, I was able to verify that the capital letters placed in B and SC(a) to organize the poems’ prosodic and syntactic structures coincide exactly with the initials — traced in yellow/gold — placed by Boccaccio in his copy of the Fragmentorum liber in MS Chigiano L.v.176 (with the only exclusion of the subdivisions of some stanzas).32 Thus we can say that in the long trajectory from ca. 1360 to the late Trecento, the majority of the stanza’s metrical subdivisions applied both by the copyists of MSS B and SC and by Boccaccio respect the original metrical scheme of the canzoni — so the coincidence does not seem necessarily striking. It is, however, remarkable to find that Boccaccio’s copy and the two late-fourteenth codices share occasionally the same but often mistaken subdivisions for the canzoni. I refer in particular to Rvf 71, 72 and 73, three canzoni 31. See Benghi 2020 for a comparison of the lectiones of these canzoni in Bodmer 131 and Sc-Ms. 93. 32. For this comparison with the Chigiano manuscript, see cc. 55v–57r in the phototype reproduction of the codex in De Robertis 1974. 122 | Textual Cultures 13.1 (2020) whose stanzas are comprised of 15 verses each and that share the same prosody and, in MS Vaticano Latino 3195, the very same mise en page, with two verses per line on seven lines, and the last verse isolated on the last line (see Magni, Storey and Walsh 2014–2020, ad loc.). The metrical scheme of the songs is aBC | bAC | CDEeD | fD | FF, therefore the correct subdivision of the stanzas would be in 3+3+5+2+2 verses. And yet Boccaccio, for all three canzoni, applies the marking initial to vv. 1, 4, 7, 10 and 14, thus mistakenly dividing the stanzas in 3+3+3+4+2 verses. MSS B and SC, for their part, show a subdivision in 6+3+4+2 verses in Rvf 71 and 72, which clearly replicates Boccaccio’s mistaken subdivision of the sirma, and then a subdivision of 3+3+3+4+2 verses for the stanzas of the canzone Poi che per mio destino, Rvf 73, just as Boccaccio had transcribed the poem. It is noteworthy that scholars have identified in Petrarch’s authorized copy in Vatican Latino 3195 four small paragraph markers that are placed and subsequently erased at the beginning of the intra-strophic units of the first stanza in each of the three canzoni, a device that Petrarch, his copyist Malpaghini or a subsequent reader might have inserted in order to ensure the accurate understanding of his challenging prosody.33 As in the course of other research I have found that B and SC partially derive from the same manuscript tradition from which derives also the Chigi copy,34 it may not be too daring to suppose that graphical solutions and metrical errors similar to the ones we have seen in Bodmer 131 and Sc-Ms. 93 were already practiced in ‘non-authorized’ copies of the Fragmenta circulating in a time preceding the spread of the model of Petrarch’s own copy in the holograph. Indiana University–Bloomington 33. Savoca (2008) observes that “regolari segni di paragrafo metrico intrastrofico sono davanti ai vv. 4, 7, 12, 14 per sottolineare lo schema metrico delle strofe”, while also admitting that “di questi, poco visibili ad occhio nudo, tre (non quello davanti al v. 7) sono segnalati da Modigliani [1904] come rasure” (113). The presence of these marks would be worth additional study. For Modigliani’s paleographical notes on each of the canzoni, see Modigliani 1904, ad loc., now reprinted in Belloni, Brugnolo, Storey and Zamponi 2004, 256, 258 and 259. For the digitial reproduction, see Magni, Storey and Walsh 2014–2020, cc. 15v, 16v and 17r. 34. In Benghi 2020. On the crucial topic of the “Chigi tradition” see Foresti 1927/1930, Seren Schoepflin 2000, Salvatore 2014 and Storey 2015. G. Benghi : Transcribing Petrarch’s Genres in the Late Fourteenth Century | 123

"As we think about the possibility of literary history and about the meaning of textual authority in relation to Petrarch, it will be increasingly necessary to take into account as many of the scribal versions of his poems as we can" (Del Puppo 2004, 131). With these words, Dario Del Puppo was sending an open invitation to further proceed on the inspection of the "evolving transcriptional pragmatics" of Petrarch's poems in the fifteenth century (2004,130). MSS Bodmer 131 from the Fondation Martin Bodmer of Cologny (Geneva) and Sc-Ms. 93 from the Biblioteca civica Gambalunga in Rimini are two copies of Petrarch's Rerum vulgarium fragmenta produced in the Veneto in the late fourteenth century that allow us not only to broaden the spectrum of "witnesses of how he [Petrarch] was read and understood" (Del Puppo 2004, 130), but also to redate to the last Textual Cultures 13.1 (2020): 106-127. DOI: 10.14434/textual.v13i1.30074 decade of the Trecento the time of that widespread circulation of Petrarch's lyric poetry when scribes undertook a variety of editorial choices, altering the poet's unique "visual poetics" within just a couple of decades from his death. 1 The two manuscripts have been dated from the end of the fourteenth century in their respective catalogues, and, in agreement with the given dates, I intend to demonstrate the importance of dating these two codices by the end of the fourteenth century, especially for their copyists' editorial choices in altering Petrarch's poetics. 2 The anonymous scribes of these two copies, in fact, align the text in the single column, one-verse per line format that "took root at the end of the Trecento", a transcriptional style that became the norm of "the vulgata in the Quattrocento" (Del Puppo 2004, 128). The popularity of this, so called 'modern', layout in the fifteenth century is renowned, while it might have been underestimated by literary historians for the fourteenth century. In the codicological examination made by Marco Cursi (2014) of the earliest manuscripts of the Canzoniere (Rerum vulgarium fragmenta, or the Rvf) we notice that the group of codices with layout "ad una colonna" is made up of only six manuscripts out of the total of 29 dated as "antiquiores", and that only three out of those six are datable to the end of the fourteenth century. 3 However, our MSS Bodmer 131 and Sc-Ms. 93 can be rightly added to the group, as they fully respect the two criteria applied by Cursi to his examination -the two criteria being 1) a dating within the first decade of the fifteenth century, and 2) the nature of the collections as books, and not as miscellanies. 4 By adding Bodmer 1. The term "visual poetics", coined in the 1980s by Storey (1989;see also 1993) and now in wide usage, refers to the use of transcriptional and typeset layouts in manuscripts and printed works as part of the intentional poetics of a literary composition. 2. For MS Bodmer 131, a detailed codicological description by Allegretti 2003 and the digital version of the codex are available on the website of the "Virtual Manuscript Library of Switzerland": https://www.e-codices.unifr.ch/it/description/fmb/cb-0131/. For Sc-Ms. 93, instead, I refer to the two descriptions available on the website of the "Biblioteca Digitale Italiana", by selecting "Biblioteca Civica Gambalunga" and searching the codex with its previous shelf-mark, that is D. II. 21, in both "Catalogo Lucchesi" and "Catalogo Nardi": http://cataloghistorici.bdi.sbn.it/indice_cataloghi.php. See also the Images in my Appendix for . See the "Tabella 5 -Codici dei Rvf ad una colonna" in Cursi 2014, 244. 4. Cursi (2014 suggests that "ammesse nel corpus le raccolte palesemente orientate verso l'adesione ad un modello di libro lirico d'autore, pure se aperte a contributi 131 and Sc-Ms. 93 to this group of codices of the Fragmenta as classified by Cursi, the number of manuscripts in one-column layout and dated to the end of the Trecento increases from three to five. When we compare the other charts made by Cursi with all the types of layouts employed by the codices antiquiores, but confining the analysis to the last quarter of the fourteenth century, we realize that the number of codices with one-column layout (5) is almost equal to the number of codices with other types of layouts: the manuscripts with a two-column layout but with the verses copied in vertical order rather than horizontal ("a impaginazione verticale", Cursi 2014, 242) are also 5 (6 when including MS London British Library, King's 321 dated "anno 1400" by its Venetian copyist), while the number of codices employing a two-column layout but with a variety of copying solutions between horizontal and vertical readings ("a impaginazione mista", Cursi 2014, 240) are 3, or 4 if including MS Firenze Biblioteca Laurenziana Segniano 1, dated to between the end of the fourteenth and the early fifteenth century first by Storey (2006, 295) and later again by Cursi (2014, 240 Thus, while the use of the two-column layout shows to be prevalent overall in this early period, when we focus on the variety of changes to Petrarch's 'noble' mise en page already employed by the scribes at the end of the fourteenth century, we realize that the one-column layout was not simply the outcome of a progressive corruption of Petrarch's visual poetics, but rather part of a spectrum of editorial choices available to the scribes, behind whose choices there were specific reasons caused by the circumstances of the copying process and/or imposed by the terms of the manuscript's commission: a situation well described by Storey with reference to the first half of the Quattrocento, 6 but that may be already happening by extravaganti, escludendo invece i testimoni riconducibili alla tipologia del canzoniere miscellaneo", and limits the manuscript's dates "dagli anni in cui l'autore era ancora vivente fino al primo decennio del secolo XV" (229). 5. For a prolonged description of the Petrarch's two-column layout and its philological and interpretative implications in his visual poetics, see Storey 1993, 201-419. 6. "una diffusione del Canzoniere multiforme già nella prima metà del Quattrocento, ormai inestricabilmente legata alla variabile di copia e alle scelte 'edi-the last decade of the Trecento, for we see in the comparison above that some of the eminent copies of the Rvf, such as MS Laurenziano Pluteo 41.10 and Paris Bibliothèque nationale de France, Italien 551, date from slightly later than other manuscripts with the one-column layout like the Bodmerian and Riminese codices. 7 The evidence offered by witnesses such as MSS Bodmer 131 and Sc-Ms. 93 can therefore address this research along the paths indicated by Brugnolo,Storey,Del Puppo,and other scholars,8 which is to inspect the formats of the early copies of the Fragmenta in order to verify what they can tell us about their antegraphs and the earliest circulation of Petrarch's poems. As Del Puppo points out, "Wilkins does not discuss whether the poems in earlier forms of the Canzoniere are laid out as they are in the holograph, Vaticano latino 3195, and whether scribes would have followed these formats" (2004,116), while this is, in fact, a crucial point that awaits further investigation.
Before proceeding, I must outline the unique list of features that Bodmer 131 (for which I will use the initial "B") and Sc-Ms. 93 (SC) have in common. Codicologically, they depict the favorite forma-libro, or book form, of the earliest tradition of the Canzoniere, which is an average-sized book with gothic or semi-gothic script on parchment. 9 By coincidence, they are both composed of two different units: in B, the first section (cc. 8r-143v) contains the Rerum vulgarium fragmenta plus 36 mostly interspersed Distoriali' dei vari copisti, siano esse frutto di adattamenti (o fraintendimenti) di comodo del singolo amanuense o il riflesso di una nuova esigenza socio-culturale dettata dalla committenza", in Storey 2006, 300. 7. While Del Puppo describes MS Laurenziano Pluteo 41.10 as "made during the last two decades of the fourteenth century in the Veneto" (2004, 121), Cursi suggests that the manuscript was produced sometime between 1397 and the first years of the Quattrocento (Pulsoni -Cursi 2010, 258-68); Teresa De Robertis recently dated it "circa 1400" (2016, 67). 8. See especially Brugnolo 1991, Pacioni 2004, and Del Puppo 2004and 2007. Cursi generally describes the standard format for lyric transcriptions: "per le Rime il modello librario preferito prevedeva il materiale scrittorio membranaceo, misure medie, scritture gotiche o semigotiche" (2014,233). MS Bodmer 131 measures 230 x 163 mm, with a writing space of 32-33 lines, while Sc-Ms. 93 is 210 x 113 mm, with a ruled writing space varying between 24 and 26 lines in the first part and 23 in the second part. As for the script, it is a littera textualis; Pancheri (1993, 78) claims that it has a touch of bastarde.
perse (or Estravaganti), 10 while the second section (cc. 144r-178r) contains 29 poems by Dante plus seven by Guinizzelli, Cavalcanti and Cino; in spite of its position at the end of the book, the script and the construction of the quires suggest that the second section was transcribed earlier than the first one. 11 On the other hand, what undermines the unity of MS SC is not the content but an abrupt change in the script and in the mise en page, on c. 61v (in the midst of Rvf 181, v. 4, see my Appendix Image 1 and Image 2), due to the intervention of a second scribe, so that the manuscript must be considered as two units, independent from one another both in their codicological and textual aspects; these two units I will indicate as "SC(a)" and "SC(b)". A significant common trait between B and SC, instead, is the identical sequence of the poems -though only up to Rvf 215, and excluding the mechanical lacunae both at the beginning, at the end, and in the midst of the collection. 12 In addition to containing the same four-10. "Disperse" (uncollected) and "estravaganti" are the terms usually used for those poems not included by Petrarch in his holograph (MS Vatican Latino 3195) but attributed to Petrarch by the manuscript tradition. The first and only complete edition of all these spurious poems was edited by Angelo Solerti in 1909, and recently republished in fac-simile with an introduction by Vittore Branca and an Afterword by Paola Vecchi Galli (see Solerti 1909 andVecchi Galli 1997). 11. I provide a more detailed analysis of the Dante and Stilnovo section of MS Bodmer 131 (B) in Benghi 2020. For the convenience of online and remote readers, references to the Bodmer manuscript will be to the most immediate access to the codex in its digital form in the e-codices site (see n.2 above). This means that in some cases the original numbering of the chartae in the upper right-hand corner of the recto of each charta will not be reflected in the digital access information for the manuscript due in part to a missing charta between the index (cc. 1r-7v in the original numbering) and its initial leaf (c. 9r in its original numbering = digital c. 8r [note the reference on c. 7v in the ancient index to Voi che ascoltati in rime sparse il sono [sic] as c. 9]). Those who consult the codex directly in Cologny should adjust the numbering accordingly. 12. See the Appendix below with the complete sequence of the Rvf in both manuscripts. Please note that the numbers assigned correspond to the order revised by Petrarch by means of marginal numbers in his holograph MS Vatican Latino 3195. Thus my numbers are different from those of the physical order in which the last 31 poems appear in the manuscript, the order that was adopted by Wilkins (1951) and in the Petrarchive (Magni, Storey andWalsh, 2014-2020), the latter of which indicates Petrarch's revised order as "rev". Thus the final sequence of B, described by Wilkins (1951, 232) for what he knew as "Melziano A" (the codex resided in the Melzi Library in Milano at the beginning of the twentieth century), was: "356-360, Un clima, c, cxxix, 361, 366" is in its teen Disperse copied in the same position, the two manuscripts share -in fact -the same inversions in the two sequences Rvf 90, 89, 91 and Rvf 153, 155-158, 154. 13 Finally, a comparison of the lectiones reveals that B and SC share numerous variants (varianti congiuntive), while also containing a minority of different readings (varianti disgiuntive); therefore, we deduce that they must descend, even if not directly, from the same antegraph (as documented also by their sequence), while we must exclude that one is copy of the other. An examination of their mise en page confirms this close relationship between the two manuscripts, which, based on an analysis of the variants, I have identified as similar to two 'cousins': they share the same system of graphic features, and sometimes even the same mistakes, to reproduce the metrical schemes of the Fragmenta, proving that their formats derive mostly from their antegraph, and yet the two copies also contain independent mistakes by either one of the scribes. 14 All this is true, however, only pertaining to the first section of SC, that includes Rvf [*44]-181 (Rvf 1-43 are missing due to mechanical loss of the chartae), since the change of scribe on c. 61v coincides with a change of the antegraph -as a change in the lectiones also confirms. 15 MS Bodmer 131 is a very clean, finished product, which displays lavish illuminations of leafage and dragons on its opening charta (c. 8r), and has each poem initial in red ink with blue entrelacs, while all capital letters are marked in red. The Riminese codex, in its first section, is instead completely devoid of all rubrics and decoration, leaving empty spaces for the missing decorated initials (of the size of three or four writing lines, as in B). In spite of the different aesthetics, the scribes of both B and SC(a) organize all five genres of the Canzoniere in the same one-column, one verse per line format, and distinguish the five poetic genres of the Fragmenta through the use of capital letters, that are always located slightly toward the left margin (and that, in B only, are marked in red), in order to separate the metrical units of each genre. It is useful for the modern reader to remember that these capital letters do not follow a period, and therefore do not express the real syntactic flow of the discourse, but serve purely as metrical markers These are the fourteen Disperse contained both in B and SC-MS 93: CCXIII, XXV, CC, CLXXXII, XXVII, XXIX, XIX, LXIV, CX, XLVIII, L, C, CXXIX. 14. The nomenclature suggests that they are copies of two different exemplars that, in turn, derive from the same antegraph. 15. For an extended examination of this topic, see Benghi 2020. (in any case, in B and SC, punctuation is rare, and mostly represented by the punctus).
Among the five genres, the sonnet stands out for its typical, unchanging structure of 4+4+3+3 verses, resulting from the identification of the quatrains and the tercets by the capital letter that introduces the first verse of each metrical unit. As modern readers, we tend to take this layout for granted, failing to realize that in this way the constituent unity and visual solidity on which Petrarch's sonnet was conceived is completely disrupted, 16 for "Petrarch's sonnet format is to assure integration of the tercets into the body of the sonnet by not spatially distinguishing them from the octave" (Storey 1993, 238). A brief digression needs to be opened here to recall that the breakage of the sonnet's unity as designed by Petrarch is already found in some of the 'faithful' copies such as MS Laurenziano Pluteo 41.17, where a paragraph mark is placed next to the first tercet, or in MS Laurenziano Segniano 1, where the capital letter of v. 9 (that is the beginning of the first tercet) is taller than all the other versali, occupying two writing lines. Not only the separation of the tercets from the octave was common practice, deriving directly from the poetic tradition of the thirteen century, 17 and applied independently from the type of layout (for instance, in MS Trivulziano 1091, that has a mixed layout, 18 the capital letter of v. 9 is marked in red 19 ), but also the breakdown of the tercets seemed somehow 16. "La forma progettata del cod. Vat. lat. 3195 è infatti concepita per essere visivamente e semanticamente organica, unitaria e compatta nel suo svolgimento in progressione dall'ottava (le due quartine) alla prima terzina e soprattutto da questa alla seconda terzina", in Storey 2004, 153. 17. As Storey explains, "l'impostazione petrarchesca riprende modalità di mise en page codificate dal duecentesco cod. Vat. Lat. 3793, tralasciandone i marcatori paratestuali più superficiali, quali le graffe per le distinctiones più difficili" (2004,153). As a matter of fact, the scribes of MS Vatican Latino 3793 tend to place a paragraph mark next to the first tercet, and also emphasize the overall structure of the sonnet as composed of an octave plus two tercets by tracing two brackets on the right side of the sonnets, with the first bracket enclosing the two quatrains (vv. 1-8) and with the second bracket enclosing the two tercets (vv. 9-14). 18. "ad impaginazione mista" (Cursi 2014, 240). This codex can be seen on line, thanks to the "Manus Online" project: https://manus.iccu.sbn.it//opac_Sche-daScheda.php?ID=50126. It contains Petrarch's Fragmenta and also the fifteen canzoni distese, as we find in MS Bodmer 131. But in MS Trivulziano 1091 Dante's songs are at the beginning of the collection, before the Rvf. 19. Starting from c. 10v, in Rvf 12. In this sonnet, we actually see a red paragraph mark next to v. 9, presumably to point out even more clearly the beginning of necessary to the eye of most readers -as Boccaccio's renowned copy of the early version of the Fragmenta, the Fragmentorum liber, in MS Vatican Chigiano L.v.176 witnesses, with the regular red and blue paragraph marks that introduce respectively the first and the second tercet of each sonnet. As we know, Boccaccio's copy dates from at least twenty-five years earlier than the other ancient manuscripts so far taken into account (and also embodies the gothic culture Petrarch strived to reform with his revolutionary visual poetics), therefore it might seem an inappropriate example. 20 However, when we look again at MS Trivulziano 1091, dated 1391-1410 by the catalogue of the Trivulziana Library, we find that the copyist felt the need to separate the second tercet from the first one by placing two oblique dashes between vv. 11 and 12: in other words, "il copista applica una 'traduzione culturale' per cui impiega una rete -a volte anche microscopica -di meccanismi per rendere la veste grafica del sonetto il più leggibile possibile per un pubblico attuale" (Storey 2017, 97 Going back to the graphological organization of the genres in B and SC(a), ballate and madrigali are also identified in their metrical schemes by capital letters, or initials, so that they are distinguished from the sonnets by their variable units (and length), reproducing rather faithfully Petrarch's metrical experimentation. In B, for example, it is striking to find that the copyist doesn't fall into the error of confusing the two ballate grandi he does not pay attention to the separation of two piedi from the volta. In Rvf 59 (Perché quel che mi trasse ad amar prima, c. 31r), another ballata mezzana with the same scheme of Rvf 55, the copyist instead begins v. 4 with capital letter, and then again v. 6, 11 and 15, thus establishing the scheme 3+2+5+4+3, which blurs the original formula of the two piedi (2 verses+2 verses) followed by the volta (3 verses), repeated twice. 21 Unfortunately, due to the loss of several chartae in SC, we cannot complete a comparison of these ballate. However, SC(a) does contain the two remaining ballate grandi, Rvf 63 and 149 (Volgendo gli occhi al mio novo colore, cc. 51r-v, and Di tempo in tempo mi si fa men dura, c. 70v), that share with B the same (correct) scheme, though in B we find a red paragraph mark next to v. 5 (i.e. the beginning of the piede after the refrain) of Rvf 149, that does not appear in SC(a). 22 The case is particular, since neither in B nor in SC(a) did the scribes ever apply the paragraph marker to the genre of the ballata, while we remember that Petrarch uses it in the three ballate Rvf 55, 59 and 63, but not for Rvf 149. 23 Moving to the four madrigals of the Rvf, the division of the verses substantially resembles the original metrical schemes: The only exception occurs in B's transcription of Rvf 54 in which the copyist fails to distinguish the two couplets. 24 The division units, indicated in brackets above, are as follows: 3+3+2 for Rvf 52 (c. 28r in B), 3+3+4 for Rvf 21. The metrical scheme of this ballata mezzana is as follows: XyY Ab Ab ByY Cd Cd DyY. 22. The breakdown of the verses in Rvf 63 is 4+3+3+4, which represents the metrical scheme XYYX ABC BAC CDDZ, and in Rvf 149 is 4+4+4+4, representing X(x)YyX AbbC AbbC CDdZ. 23. We should note, however, that Rvf 149 is the only ballata that employs mid-verse rhymes, rima al mezzo, in the refrain, an element that might play a role in this graphic solution by the copyist of B. We should also keep in mind that these poems were actually copied by Malpaghini (see Storey 1993, 264 and2004, 158), whose work Petrarch oversaw and corrected.  1-3 and 4-6), now distributed over three lines of transcription. While Petrarch's own mise en page for Rvf 52 (c. 11v), Rvf 54 (c. 12v) and Rvf 121 (c. 26r) reinforces the prosodic and syntactic phrasing in these madrigals' tercet structure, the layout of Rvf 106 emphasizes the meter and closure of the interlocking rhymes -orta (vv. 1, 4 [accorta -scorta]) and -ino (vv. 3, 6 [destino -camino]) that link internally the syntactic construction of the first six verses against the closing couplet (vv. 7-8 [poi -suoi]). 25 In the case of the Bodmer copy, the scribe's transcriptional system ignores the integrative value of Petrarch's approved layout designed to underscore the madrigal's variation. The 3+5 visual structuring of the madrigal in B instead reinforces the madrigal's opening tercet, and, in opposition to Petrarch's model, links the second tercet to the final couplet (see Fig. 2).
The genre of the canzone in the Bodmer codex demonstrates instead two-line decorated initials that introduce every stanza, while the internal structure of the stanzas is conveyed by regular-sized and rubricated small initials. In essence, these decorated initials opening the stanzas actually distinguish the canzoni from the other genres, as opposed to the red paragraph marks employed for the stanzas of the sestina. This distinction is only true in B, since in SC(a) the sestina is simply organized by small initials at the beginning of each stanza. It is remarkable that in both copies the sestine, calculated by Petrarch as part of the canzone genre, are transcribed without errors, with their six stanzas succeeding one another in the correct order. 26 We find one exception in SC(b) (cc. 120r-121r): the verses of Rvf 332 (Mia benigna fortuna) -the only double sestina (sestina doppia) of the Fragmenta -are copied in a strangely alternating sequence, that is 1, 24, 2, 25, 3, 26, 4, 27, and so on. This is a significant error for which a clear understanding of B's and SC's common ancestor can guide us. What has happened here is that the scribe of SC(b) has copied from an antegraph 25. Marco Pacioni (2004) remarks that even in Petrarch's short madrigals variation in rhyme constructions are in invariably reflected in a nuanced mise en page ("la variatio rimica è così strutturante che è necessario renderla sempre evidente sul piano grafico, nella mise en page" ). 26. One error is found in B's copy of Rvf 239 (c. 93v), in which the scribe fails to copy the congedo, which is then added by the same hand on the left margin of the poem, thus this case is to be considered simply a distraction of the copyist rather than due to the complexity of Petrarch's original layout. whose transcription of the double sestina is in the same two-column mise en page as that of Petrarch's authorized copy, MS Vatican Latino 3195, which lays out the sestina, uniquely among the five genres, in a vertical reading down the left-hand column and then continues the song's remaining verses down the right-hand column, invariably producing two columns of unequal length with a blank space at the bottom of the right-hand column. 27 This change in Petrarch's usual horizontal layout, from the left to the right column for all genres except the sestina, clearly caused some confusion with the SC copyist, just as it did with the scribe of Laurenziano Segniano 1. 28 The copyist of SC read his exemplar from the left to the right column, just as he had clearly done in the case of the other poetic genres, converting their horizontal reading strategy into a vertical presentation. We should note one additional detail in SC: red paragraph markers are placed next to vv. 25, 7 and 31, which are the first verses of, respectively, the fifth, second, and sixth stanzas. When the copyist returns to rubricate his copy -for the rubricator's hand coincides with the scribe's -he identifies the correct beginnings of every stanza in spite of the wrong order of the verses. The use of paragraph marks in the sestina Rvf 332 is even more puzzling when we realize that this is the only case in which the rubricator applies them to a sestina (rubricated as "sonetto"), and that the only two additional places where the same red markings are found are not in sestine but in the canzoni S'i'l dissi mai ), rubricated as a "sonetto", and Ben mi credea ) -rubricated as "cançona morale". MS Bodmer 131 instead reveals an example of confusion between the genres of canzone and sestina, a problem originating perhaps with the disruption of Petrarch's mise en page: on cc. 18v-19v, Rvf 29, Verdi panni, a canzone, which is punctuated with red paragraph markers reserved for the sestina next to each stanza, just like the poem that follows it, Rvf 30, Giovane donna, which is a sestina. But the unique prosody of Rvf 29, which consists of coblas unissonans, requires the same rhymes for each stanza, a repetition that might have caught the eye of the rubricator, leading him to mistake the canzone for a sestina (though the copyist doesn't seem to mind that each stanza of the mistaken sestina has seven verses rather than the sestina's required six).
As we recall, the "struttura metrico-strofica" (Storey 2004, 162) of the canzoni is conveyed as well by the rubricated initials, by which the metrical sub-units of each stanza are identified, and, overall, faithfully represented. And yet, in addition to the loss of Petrarch's subtle experimentation in 27. For a description of this standardized use in Petrarch's visual poetics, defined by Storey as "descriptive sestina space", see "Sestina" and "Spaces" in the Glossary of Magni, Storey and Walsh 2014-2020, http://dcl.slis.indiana.edu/petrarchive/content/glossary.xml. 28. The copyist of MS Laurenziano Segniano 1, from the first part of the Quattrocento, standardizes all the layouts of Petrarch's lyrics in the Rvf to conform to Petrarch's horizontal reading strategy, also applied to the sestinas. variation (variatio) based on similar metrical structures, copyists required to intervene between the two systems of the copy and his modern reader invariably misunderstood certain forms. In Rvf 127 (In quella parte dove Amor mi sprona), for instance, the copyists of both B (c. 59v-61v) and SC(a) (cc. 38v-40v) structure the first three stanzas with initials to create four internal units of 3+3+5+3 verses for the rhyme scheme: ABC BAC CDE eDe FF. 29 But from the fourth stanza on the subdivision changes to 3+3+3+3+2, and then in the fifth stanza it changes again into 3+3+3+2+3 perhaps to reflect the changing syntactic structures of the stanzas. In the famous canzone Italia mia, benché 'l parlar sia indarno (Rvf 128, cc. 61v-63v in B, cc. 40v-43r in SC), we notice even greater variation in how the copyists organize the metrical structure of each stanza whose rhyme scheme is AbC BaC cDE eDdf GfG. The first stanza is divided into units of 3+3+3+4+3 verses, the second in 3+3+3+2+2+3, the third like the first, the fourth is instead changed to 3+3+2+6+2, the fifth changes again to 3+3+3+2+3+2, and the sixth to 3+3+3+3+2+2 and 3+3+3+5+2 for the seventh and final stanza. It is noteworthy is that these variations are identical in both B and SC(a), as it happens for many other poems as well, such as Rvf 71, Perché la vita è breve, for which the copyists change the repeated 6+3+4+2 visual-syntactic division of the verses to 5+3+5+2 for the last stanza, or Rvf 72, Gentil mia donna, i' veggio, where they change the same structuring division (6+3+4+2) into 3+3+3+6 for the last stanza. In Rvf 129, Di pensier in pensier, di monte in monte, along with the same changes in the subdivision of the stanzas, the two copies also share an error in the transcription, introducing the third stanza with a larger initial, as if it were the incipit of a new song. In fact, the copyist of SC(a) even leaves a line of blank space before this stanza. Yet only in SC(a) the fourth stanza is also introduced by a taller, three-line initial (instead of two), thus causing the stanzas of Di pensier in pensier (Rvf 129) to be split into individual poems (compare c. 64r in B with cc. 43v-44r in SC(a), see Appendix, Image 3). This and other occurrences of the same errors (congiuntivi) in the transcription (often sustained by the same variants in the text) are, at the same time, accompanied by independent errors (disgiuntivi), leading us to exclude the possibility that one manuscript may be a copy of the other. Rvf 126 offers, in this regard, one more excellent example: it contains one variation of the strophic division in common, and one instead unique to SC(a); moreover, the copyist of B omits v. 56, and adds it in the margin (c. 59v), while in SC(a) it is transcribed regularly in the body of the text (c. 38r); but in both copies the verse reads "orsi charco doblio", differing from Petrarch's holograph ("cosi carco doblio").
The second section of SC (to which I refer as SC(b)), nonetheless, must derive from another antegraph, proved by the readings and its very different modality of transcription: all the poems, still in a single-column layout, are copied with the head verse (versale) opening each verse (always traced in red), without deference to the prosody, and therefore with no distinction between genres. Moreover, all the poems are separated only by a oneline rubric (in the same hand) that reads "Sonetto di mess(er) franc(esc)o petrarcha" for every genre, whether it is actually a sonnet, short poem (as madrigals and ballate were sometimes deemed) or not, as it happens for the sestina Rvf 214 (on c. 74v), for all the ballate and madrigals, and for several canzoni, while there are only a few canzoni that carry the rubric "Cançona morale". 30 In the homogeneous and compact layout of SC(b), visually the opposite of the division of the metrical units of SC(a) and B, the few graphological devices employed to organize the prosody stand out, along with the inconsistency with which they are used. Rvf 206 and 207 (S'i' 'l dissi mai, ch'i' vegna in odio a quella and Ben mi credea passar mio tempo omai, cc. 69v-72v, the first two canzoni in our section SC(b)) are transcribed in a prose format, and with paragraph markers inserted only to signal the stanzas. By the same token, the same copyist transcribes Rvf 264,268 and 270 (I' vo pensando,et nel penser m'assale,Che debb'io far? che mi consigli,Amore? and Amor,se vuo' ch'i' torni al giogo anticho,c. 90r to 97v) rigorously one verse per transcriptional line, with the stanzas distinguished by the empty space for what we suppose to be decorated initials. Even more variations occur in subsequent canzoni: in Rvf 323 (Standomi un giorno solo a la fenestra, cc. 113r-114v) there is no separation of the stanzas whatsoever; in Rvf 325 30. We should keep in mind that it is common in many copies of the end of the fourteenth century to find a discrepancy between the differentiation of the genres by layout and graphic devices and their actual labeling: Petrarch's poems are usually referred to, in fact, as either "sonetto" or "canzone" (usually accompanied by "morale"), as we read in the rubric opening the Rvf in MS Trivulziano 1091: "Sonetti echançoni morali di mess(er) francescho P(etrarca) poeta fiorentino" (c. 9v). It is interesting to find an individual label of the genre of the madrigale (expressed by the letter "m." in opposition to "s." and "c." of all the other poems) in the index of MS Trivulziano 1015 (dated last quarter of the fourteenth by Cursi 2014, and accessible online through the Manus database: https://manus.iccu.sbn.it/opac_SchedaScheda.php?ID=229873).
(Tacer non posso, et temo non adopre, cc. 115r-117r) the scribe leaves one empty line between the stanzas, and yet he has copied Rvf 325 immediately after the ballata Rvf 324, Amor, quando fioria, without an introductory rubric nor any space between the two poems. What determines these changes in copying the same genre? Was the scribe of SC(b) simply negligent, or was he perhaps sensitive to subtle variation? Can we conjecture that he was copying from a single antegraph for Rvf 206-207, from a different exemplar for the group Rvf 264, 268 and 270, and then from a third (and even forth) for the next songs? It is not very likely that a single copyist would have consulted three or four different exemplars to reconstruct a text like the Rvf. 31 Rather it is more likely that some canzoni were read in a different way than others, and therefore reproduced differently. In this light, the reason why only a few songs were introduced by "Cançona morale", instead of by "Sonetto" as the majority, could find explanation, though the -mostly wrong -identification of the genre of the canzone still leaves us with the question on whether it was taken from the antegraph or was rather generated autonomously by the scribe. MSS Bodmer 131 and Sc-Ms. 93 show, on one hand, how Petrarch's antegraph/s must have played a key role in influencing the editorial choices of a specific format and graphological devices, while, on the other hand, these later copies' usual, one-column format and graphic-metrical system was by the last decade of the fourteenth century already being tested and solidified. Comparing all the "forme strofico-grafiche" (Storey 2004, 166) of the canzoni, I was able to verify that the capital letters placed in B and SC(a) to organize the poems' prosodic and syntactic structures coincide exactly with the initials -traced in yellow/gold -placed by Boccaccio in his copy of the Fragmentorum liber in MS Chigiano L.v.176 (with the only exclusion of the subdivisions of some stanzas). 32 Thus we can say that in the long trajectory from ca. 1360 to the late Trecento, the majority of the stanza's metrical subdivisions applied both by the copyists of MSS B and SC and by Boccaccio respect the original metrical scheme of the canzoni -so the coincidence does not seem necessarily striking. It is, however, remarkable to find that Boccaccio's copy and the two late-fourteenth codices share occasionally the same but often mistaken subdivisions for the canzoni. I refer in particular to Rvf 71, 72 and 73, three canzoni 31. See Benghi 2020 for a comparison of the lectiones of these canzoni in Bodmer 131 and Sc-Ms. 93. 32. For this comparison with the Chigiano manuscript, see cc. 55v-57r in the phototype reproduction of the codex in De Robertis 1974. whose stanzas are comprised of 15 verses each and that share the same prosody and, in MS Vaticano Latino 3195, the very same mise en page, with two verses per line on seven lines, and the last verse isolated on the last line (see Magni, Storey andWalsh 2014-2020, ad loc.). The metrical scheme of the songs is aBC | bAC | CDEeD | fD | FF, therefore the correct subdivision of the stanzas would be in 3+3+5+2+2 verses. And yet Boccaccio, for all three canzoni, applies the marking initial to vv. 1, 4, 7, 10 and 14, thus mistakenly dividing the stanzas in 3+3+3+4+2 verses. MSS B and SC, for their part, show a subdivision in 6+3+4+2 verses in Rvf 71 and 72, which clearly replicates Boccaccio's mistaken subdivision of the sirma, and then a subdivision of 3+3+3+4+2 verses for the stanzas of the canzone Poi che per mio destino, Rvf 73, just as Boccaccio had transcribed the poem. It is noteworthy that scholars have identified in Petrarch's authorized copy in Vatican Latino 3195 four small paragraph markers that are placed and subsequently erased at the beginning of the intra-strophic units of the first stanza in each of the three canzoni, a device that Petrarch, his copyist Malpaghini or a subsequent reader might have inserted in order to ensure the accurate understanding of his challenging prosody. 33 As in the course of other research I have found that B and SC partially derive from the same manuscript tradition from which derives also the Chigi copy, 34 it may not be too daring to suppose that graphical solutions and metrical errors similar to the ones we have seen in Bodmer 131 and Sc-Ms. 93 were already practiced in 'non-authorized' copies of the Fragmenta circulating in a time preceding the spread of the model of Petrarch's own copy in the holograph. Indiana University-Bloomington 33. Savoca (2008) observes that "regolari segni di paragrafo metrico intrastrofico sono davanti ai vv. 4, 7, 12, 14 per sottolineare lo schema metrico delle strofe", while also admitting that "di questi, poco visibili ad occhio nudo, tre (non quello davanti al v. 7) sono segnalati da Modigliani [1904] come rasure" (113). The presence of these marks would be worth additional study. For Modigliani's paleographical notes on each of the canzoni, see Modigliani 1904, ad loc., now reprinted in Belloni, Brugnolo, Storey andZamponi 2004, 256, 258 and259. For the digitial reproduction, see Magni, Storey andWalsh 2014-2020, cc. 15v, 16v and17r. 34. In Benghi 2020. On the crucial topic of the "Chigi tradition" see Foresti 1927/1930, Seren Schoepflin 2000, Salvatore 2014and Storey 2015 Works Cited Printed Sources