Is There a Printer’s Copy of Gian Giorgio Trissino’s Sophonisba?

This essay examines the manuscript Additional 26873 housed in the British Library. After clarifying its physical features, it then explores various hypotheses as to its origins and its intended use. Finally, the essay collates the codex with Gian Giorgio Trissino’s first edition of the Sophonisba in order to establish the tragedy’s different drafts and the role of the manuscript in the scholarly edition’s plan. The manuscript classified as Additional 26873 (Hereafter La) is housed at the British Library in London and described in the online catalogue as follows: “Ital. Vellum; xvith cent. Injured by fire. Octavo. Giovanni Giorgio Trissino: Sophonisba: a tragedy”. A few details may be added. La is part of a larger collection comprising Additional 26789–26876: eighty-eight items composed in various languages (English, French, Italian, Latin, Portuguese, Spanish), many of which have been damaged to different degrees by fire and damp. The scribe has been identified as Ludovico degli Arrighi (1475–1527), the well-known calligrapher from Vicenza (Veneto, northern-east Italy) who later turned into a typographer, active between his hometown and Rome during the first quarter of the sixteenth century.1 Arrighi’s renowned handwriting, defined as corsivo lodoviciano (Ludovico’s italic), is a calligraphic italic based on the cancelleresca, well distinguished from the latter by the use of upright capitals characterized by long ascenders/descenders with curved ends, the lack of ties between characters, the greater space between transcriptional lines, and capital letters slightly 1. Arrighi’s hand was identified by Mazzoleni 1996. On the manuscript see also Wardrop 1939 and Fairbank and Wolpe 1960. On Arrighi’s importance in handwriting and typographical history see Morison and Warde 1926; Johnson 1926, 1934, 1950; Scarfoni 1938; Ascarelli 1953; and Bonacini 1953. Textual Cultures 14.1 (2021): 106–123. DOI 10.14434/tc.v14i1.32842 D. Perotti : Is There a Printer’s Copy of Gian Giorgio Trissino’s Sophonisba? | 107 higher than those of the current cancelleresca, whereas the inclination of the letters is the same, placed between 8 and 10 degrees (see Fig. 1).2 La was not originally chartulated, nor paginated; a modern chartulation has been added in pencil in the upper right head of each charta (chartae 1–60); the sheets of each gathering are signed up until charta 4 with Latin letters together with Arabic numerals in sequence (see Figs. 2-3). 2. On Arrighi’s italics see Casamassima 1963. For a general overview related to his printings see Gaskell 1972, 20-24. Figure 1. London, British Library, shelf-mark Additional 26873, charta 19r. A detail of Arrighi’s italics. Figure 2. London, British Library, shelf-mark Additional 26873, charta 25r, signature (detail). Figure 3. London, British Library, shelf-mark Additional 26873, charta 28r, signature (detail). 108 | Textual Cultures 14.1 (2021) The format is 8o: A–G; each charta is roughly 1900 × 1300 mm, made up of nineteen poetic verses. Both the initial and final sections (chartae 1–17; 46–56) are those most damaged by fire (approximately 40–60% of the text is missing); the central section is sooty at the head of the chartae and some scraps are missing along the back, though generally speaking this section is fairly legible (see Fig. 4).3 Considering the total absence of corrections and variant readings, La is a fair copy, although fragmentary. A note on the tail of the guard, added when the manuscript came to the British Museum, states that it was “presented by J. T. Payne on 29 July 1759”. The news of a manuscript copy of the Sophonisba would not in itself be extraordinary since Trissino supervised the publication of three printed editions of the tragedy over five years: 1524 in July and again in September, both printed by Arrighi, and a third in Vicenza in 1529, printed by Ianiculo.4 However, as Arrighi was the printer of the first edition, it is easy to insist on the importance of the information derived from such a source: like an autograph for literary authors, it tells us much about Arrighi’s way of working, both as scribe and printer, and given the physical features of La, it follows logically to ask why Arrighi produced a copy of the tragedy and whether it is genetically related to Trissino’s supervised editions. Preliminarily, it is possible to hypothesize the following: 1. La is a printer’s copy for Arrighi’s first and/or second edition of the Sophonisba (henceforth labeled, respectively, Ar1 and Ar2); 2. La is a printer’s copy for an edition prior to the spelling reform of Trissino; 3. La is a copy composed by Arrighi and conceived for a restricted circulation within the Roman court of Pope Leone X (1475–1521). Let us consider the first hypothesis. Given that La is calligraphic, free of variant readings, and composed of signed gatherings, it seems at first glance to be a printer’s copy. At the same time, however, it lacks all of the typical indications of copy-preparation, such as annotations concerning 3. Due to the poor condition of the manuscript, it was not possible for me to carry out a typological analysis of the paper. I limit myself here, then, to giving the approximate height and width of the chartae. For the same reasons, nothing can be said about the original binding, since the manuscript was spun off and then reassembled following its restoration after the fire. 4. On the editions of the Sophonisba, see Davoli 2020. D. Perotti : Is There a Printer’s Copy of Gian Giorgio Trissino’s Sophonisba? | 109 Figure 4. London, British Library, shelf-mark Additional 26873, charta 16r. An example of the damages provoked by fire. 110 | Textual Cultures 14.1 (2021) layout, italicization, capitalization, ending marks of the type-pages, or even traces similar to freshly made proofs, inky thumb-marks, etc.5 Another point concerns the format: in the case of plays, format was often decided by convention, and it was quite common that the choice fell on the quarto book, though less expensive printed editions sometimes appeared in the octavo format. Both were handy formats, but the quarto was preferred, especially in the first quarter of the sixteenth century when the printed book had not yet become an entirely commercial product to be sold at low cost and when printers still aimed to provide their public with a quality product. In the case of Arrighi, a preference for the quarto can be assumed, considering that the outcome of his typography can be compared to that of a scriptorium in the quality of the paper, the page layout, and the design of the types. An accurate prevision of the length of the book was the first step in the preparation and organization of the edition, so as to come to a decision about the format and the right quantity of paper to be ordered; to this end, Arrighi or the compositor would have to cast off the manuscript copy by counting words according to the sizes of type and page that had been decided on.6 The reasons to produce an accurate copy for the typography laid in the prediction of the exact contents of each type page: with the copy casted off, setting could begin anywhere in the book, avoiding the constraint, if necessary, of following the proper page order. Hence, since Ar1 and Ar2 are in quarto with twenty-two poetic verses per charta, we can hardly suppose that Arrighi produced as his copy an octavo manuscript with nineteen poetic verses per charta, for the difficulties that would have emerged in planning the printing. The first hypothesis may also be tested by shifting the focus to the text. The comparison of La and Ar1/Ar2 reveals some important differences, the most outstanding of which concerns the failure to use Trissino’s spelling reforms. Indeed, in a historical moment in which the Italian language was experiencing a phase of important codification and standardization, Trissino proposed a reform of the Latin alphabet that aimed to implement the writing system, especially in the representation of the open/closed traits of the vowels within literary Tuscan; this is because, particularly to non5. Some authors gave clear instructions for the layout of their books, and give the great care that Arrighi reserved for the graphical aspects of his publications (both manuscript and printed), it is very likely that the fair copies he prepared included such indications, unlike La. See Greg 1966, 99–100. 6. There is no information on Arrighi’s typography to indicate whether he was himself the compositor or just the master/overseer. See Smith 1755, 273 and Moxon 1962, 192. D. Perotti : Is There a Printer’s Copy of Gian Giorgio Trissino’s Sophonisba? | 111 Tuscan speakers (the largest part of the inhabitants of the peninsula), how vowels were to be pronounced in certain phonetic contexts was far from clear due to the influence exerted by the respective dialects. Briefly, its reform included the following five additions to the current Latin alphabet:7 • ε and ω used to indicate the vowels [ɛ] and [ɔ]; • j for the semivowel [j]; • v for the consonant [v], distinguishing it from the semivowel [u]; • ç for voiced sibilant [dz]. Ɛ and ω were the two additions that made the greatest impact on contemporaries: in this way each vowel within the Italian phonetic system had a corresponding grapheme.8 What is relevant is that the Greek graphemes were used in opposition to the Latin ones: Greek for the open vowel, Latin for the closed one.9 Again, given this context, that Arrighi wrote his copy 7. Given the focus of the present essay on the material and literary analysis of the manuscript, it would be inappropriate to address the complexity and vastness of the linguistic debate of the period. On the Questione della lingua (The debate around the Italian language), there is an almost endless bibliography; see, especially, the account in Tavoni 1987–1990. Particular attention to the intertwining of literary and linguistic questions is given in Pozzi 1978, 1988, 1989. A comprehensive treatment of the linguistic history of the Italian sixteenthcentury can be found in Coletti 1993 (especially chapters i and v); Marazzini 1993 and 1994 (especially chapter xi); and Trovato 1994. 8. The other vocals of the Italian phonetic system are [a] [i] [u]. Currently the graphic system cannot distinguish between [e] [o] / [ɛ] [ɔ]; even among educated people the dialect or the italiano regionale — that is to say, the Italian spoken with significant differences in terms of phonetics, prosody and syntax among the macro-regions of Italy (northern, central and southern) — influences the way in which the open/closed traits of the vowel are employed according to its phonetic context, causing many doubts especially in formal contexts (i.e., work, public events, etc.). For the same reason, it is quite easy to understand from which region of Italy one comes from, and for the better linguists, perhaps specialized in single dialects/regions, it is also possible to distinguish between different areas/cities of origin. 9. In 1529 Trissino made changes to his spelling proposal due to criticisms raised by Tuscan literates, and, in particular, by Claudio Tolomei (1492–1596) through the Polito, which was actually an answer to the Ɛpistola (Rome, Arrighi 1524) and the Sophonisba; further signs were introduced in the new supervised edition of the Ɛpistola (Vicenza, Ianiculo 1529): ʃ for the voiced sibilant [ʒ]; lj for the lateral palatal [ʎ]; k for the voiceless occlusive prepalatal veil [k]; at the same time, 112 | Textual Cultures 14.1 (2021) without using Trissino’s additions is hardly sustainable due to the difficulties that would have emerged during the composition as well as proof-correction phases: even the ablest compositors and correctors — among whom certainly Arrighi or someone working for him — made mistakes, and the fact that the new Greek letters had to be introduced with care cannot have led Arrighi to produce a copy for the typographer without employing them. Moreover, Trissino was at the time in Rome and very likely supervised the proofreading, at least in the case of Ar2, since the correction phase was seldom left to the compositor alone. In the case of Arrighi’s Sophonisba, it is also possible that he employed a professional corrector and not, as was more usual, a piece-worker who had to pay attention to his own mistakes in his own time, and who would have been tempted to overlook such errors as they would have taken longer to mend.10 It is no coincidence that, following the release of Ar1, which generated a heated controversy among the detractors of his integrated alphabet, Trissino prevented further criticism by making a reprint (Ar2) of the tragedy only two months later. Discarding thus the first hypothesis, La could instead be a manuscript conceived by Arrighi for an edition prior to Trissino’s spelling reform. In fact, what makes us inclined towards the hypothesis of a printer’s copy is the presence of signatures on the sheets. As is well-known, the use of signatures in printing derives from the assembly of sheets in order to get them correctly oriented and in the right order. For the same reason, as the format decreases the number of chartae within the gathering increases, so that signatures were also placed on the rectos of a few leaves after the first of each sheet in order to help the binder in folding. Their presence is therefore an important fact and must be properly analyzed to draw the necessary conclusions. The signatures, both in La and Ar1/Ar2, are made up of letters from the Latin alphabet, a convention deriving from manuscript culture; manuscript culture also supplies the mechanism of the twenty-fourth folded sheet signed with the double letter (for instance ee, ff etc.), as is the a substantial modification of the previous alphabet reverses the use of ω, later employed for [o], while ε remains unaltered. This last change was introduced by Trissino according to his knowledge of the Greek language (ω corresponding to [ɔ]); but this was at the expense of the symmetry of the spelling system, which appeared to his detractors even less functional and coherent than the previous one. 10. Mistakes in individual letters always occurred in some numbers no matter how skilled the compositor; such mistakes were inevitably compounded in cases where it was necessary to add Greek letters to the Latin alphabet; see Gaskell 1972, 110–16. D. Perotti : Is There a Printer’s Copy of Gian Giorgio Trissino’s Sophonisba? | 113 case of La (charta 41r), whereas Ar1/Ar2 do not maintain this technical usage (see Fig. 5). Another difference among La, Ar1, and Ar2 is that the manuscript has signatures until the fourth sheet, distinguished by letters together with Arabic numerals, whereas letters and Roman suffixes (e.g., aij, bij) are only used in Ar1/Ar2 up to the second sheet.11 Moreover, even though La is fragmentary, charta 4r is signed a4: going backwards, charta 1r coincides with sheet a1, so that we suppose the signing of preliminary leaves was made by symbols rather than letters for the impossibility of returning the dedication to the counting;12 on the other hand, in Ar1/Ar2 the signing begins with the title-page, as the sheet introducing the dedication is signed aij (the structure is also confirmed by the register). Headlines with running titles are absent from La and Ar1/Ar2, as well as catchwords at the end of the direction line and chartulation/ pagination; in the case of Ar1/Ar2, the lack of these practices, intended to help the compositor set the pages in the right order for the printing, would have made their arrangement in the proper order very slow, so that the print run could not have been very high. Therefore, it is very likely that the edition was produced for a limited audience. This tells us a lot about the way in which Arrighi interpreted his work as scribe and typographer. Because he was first a humanist and a calligrapher, he intended the printed book as a valuable product and, consequently, he created a 11. This is due to the smaller format of La, for which it was necessary to mark more than two sheets, since each gathering was made up of eight chartae, and therefore would have created greater difficulties during the binding phase. 12. La, charta 1r, coincides with charta c3r of Ar1/Ar2, so that almost three gatherings are not coinciding and must have been signed using symbols rather than the Latin alphabet. Figure 5. London, British Library, Additional 26873, charta 41r. Detail of the double letter. 114 | Textual Cultures 14.1 (2021) sort of hybrid object, strongly influenced by the manuscript culture and at the same time open to the practices of the new medium. For that reason, his manuscripts must be treated cautiously, even though they could appear on first glance as printers’ copies. The truth is that Arrighi’s printed editions distinguished themselves from most of his contemporaries’ books in that Arrighi’s resembled more a literal reproduction of the manuscript rather than a normalized version of it with substantial changes designed to optimize the typographic work. In this respect, La represents the opposite situation: very similar to a printer’s copy, it nonetheless lacks those necessary elements which would have qualified it as such — instructions about the layout, italicization, capitalization, etc. — making it most likely to be a handwritten copy of the Sophonisba written before 1515. Indeed, we know that Trissino composed his tragedy between 1513 and 1515. During this period Morsolin (1894) reports the popularity of some of its scenes in Rome even before the premiere.13 In support of this historical data, in two letters Giovanni Salviati (1490–1553) claims that manuscript copies of the tragedy, profoundly incorrect, circulated in large numbers in the city at the time, right after Trissino finished writing it.14 The argument can be taken to further support the fact that La is antecedent to 1518 — the dating proposed by Mazzoleni — assuming it was composed precisely during this period of great fortune for the tragedy. In addition, Arrighi was already in Rome in that period and had not yet established his typographical activity (Pratesi 1962). In light of this historical and biographical data, La is more likely to be a handwritten copy produced during the Roman period in the lively circulation of the tragedy rather than a printer’s copy for an edition prior to Ar1: thus for all intents and purposes La is a scribal publication. 13. For all it is known, the Sophonisba was staged publicly for the first time only many years later in Vicenza, in the great hall of the Basilica (1562), with a stage set conceived by the architect Andrea Palladio (1508–1580), pupil of Trissino, and painted by Giovanni Antonio Fasolo (1530–1572); see Magrini 1847. 14. Giovanni Salviati, while complaining to Trissino for not sending a copy of the tragedy, however reassured him that he was aware of the difficulties of finding a scribe who would have copyed it “a suo modo” (in Trissino’s own way), which seems to be an allusion to the use of the new letters. Nonetheless, Salviati claims to have found some manuscript copies and, although they were very incorrect, had them transcribed. Salviati traced back the errors to the fact that the tragedy was copied within one night; despite that, he affirmed that many other copies were transcribed. See Trissino 1729, xvi; Salviati 1878; Morsolin 1894, 77. D. Perotti : Is There a Printer’s Copy of Gian Giorgio Trissino’s Sophonisba? | 115 In addition to this third hypothesis, a complete collation among La, Ar1, and Ar2 shows that the first witnesses an antecedent stage of the draft, with a relevant number of accidentals and substantive variants. Besides, unlike what Salviati stated regarding the flawed nature of the copies he saw during his Roman stay, La is definitely correct, reflecting the fact that Arrighi worked very accurately, both as a copyist and as a typographer. Here below is a summary of the systematic variants in the passage from La to Ar1/Ar2:15 • in La the form of the coordinating conjunction is et, whereas Ar1/ Ar2 use alternately ε/εt depending if it is followed by a word beginning with a vowel (εt) or a consonant (ε); • in La the articulated prepositions are written in analytical form, while the synthetic one is used in Ar1/Ar2 (a ’l → al; de ’l → del etc.); • as regards punctuation, a systematic intervention appears at the end of the poetic verses: where in La there is a full stop or two points, Ar1/Ar2 substitute them with the semicolon; further, in La the circumflex accent is used to point out the stressed syllable of the past tense (passato remoto), whereas in Ar1/Ar2 is the acute accent (andô → andὼ; formô → formὼ etc.); occasionally, La accentuates completely independently from Ar1/Ar2, such as in charta 27v, 16: Si puo ʃperâr che ʃi ritorni àl bene → Si puὼ sperar che ʃi ritorni al bene; • in La the form of theonyms is inconstant, alternating the upper and lower case, whereas Ar1/Ar2 regularly maintain the upper case (adio → a Dio; fortuna → Fortuna; iove → Giωve); as an isolated counterexample Giove → Giωve, which not only alternates upper and lower case, but also the previous Latinism (iove) with current Italian; • the treatment of elision: in Ar1/Ar2 there is a general tendency towards elision reserved for metric needs, whereas in La the elision is almost constant with few exceptions: ch’el → che il; che’nfin → che infin; Por l’offeʃe → Por le offeʃe; tropp’obligai → trωppo obligai; as counterexamples due to metric reasons: che ogni → ch’ωgni; entro a la → entr’a la; tanto è → tant’ὲ; queʃta è → queʃt’ὲ → voʃtra ira → vωʃtr’ira;16 15. On the left the reading of La, on the right that of Ar1/Ar2. 16. It is worth underscoring that in La the elision regards above all the syntagms article + noun or relative pronoun + article. In the passage between Ar1 and 116 | Textual Cultures 14.1 (2021) • the verbs ʃopportare, domandare, promettere, eʃʃere, aʃpettare, avere are treated differently: as regards ʃopportare and domandare, a vowel shift concerns the stressed syllable (o → i ; u → o): domandô → dimandὼ; ʃupportar → ʃopportar; the verb promettere is differently conjugated in the past form (passato remoto): promiʃi → promeʃʃi; the same happens with the conjugation of passive verbal forms: ponanʃi → ponganʃi; • when part of the stressed syllable, the etymological Latin diphthong is not constantly maintained both in La and Ar1/Ar2: poʃe → puoʃe; tuor → tωr; fuori → fuωre; intieri → interi; muora → mωra; huomini → hωmini;17 • the diacritic -i is absent in La: guance → guancie; • in La the Latin proton -i is sometimes retained, sometimes not, whereas in Ar1/Ar2 its use is constant (I → ie): giovinile → giovienile; domandarne → dimandarne; on the other hand, it is not always easy to judge whether the proton -u is deriving from Latin or Venetian dialect, since the passage O → u in the Lombard and Venetian dialects (particularly the Vicenza and Padua dialects) often responds to a dialectal -u, which likewise could be ascribed to Trissino’s linguistic reform, markedly multilingual; the passage La→ Ar1/Ar2 is always o → u; • different forms for the present tense of the verb andare (firs person, singular): vvo → vὼ; • consonantal shift (voiced → voiceless) in the internal group -gr → -cr: lagrime → lacrime; • vocal shifts (i → e; o → u; e → i o → i): intrô → entrὼ; ingenocchiata → inginocchiata; devotamente → divotamente; ʃomigli → ʃimigli; ʃecura → ʃicura; • in La the Latin consonantal nexus -CZ is regular throughout the text, whereas Ar1/Ar2 alternates it with -z/-zz: gentilecza → gentilezza; nocze → nωze; dolcecza → dolceza; allegrecza → allegreza; • gemination is more consistent in Ar1/Ar2, with few exceptions: legittima → leggittima; ritrovamo → ritrovammo; debia → dεbbia; rubata → rubbata; fanciulo → fanciullo; Ar2, the trend towards a widespread elision is more marked, as reported; see Davoli 2020, 17. 17. In those years diphthongization was in the literary language assuming its regularized form but was still not predominant. D. Perotti : Is There a Printer’s Copy of Gian Giorgio Trissino’s Sophonisba? | 117 • La is less consistent in pseudo-etymological spellings if compared to Ar1/Ar2: ancora/ancor → anchora/anchor; qualcuno/qualcun → qualchuno/qualchun; some are retained or restored in La: chara → cara; loco → luωgo; palagio → palazo; • La is inconsistent in the treatment of the Latin ending -TIA: patienza → patiεntia; reʃiʃtenza → reʃiʃtεntia; temperanza → temperantia; but also: ʃententia → ʃentεntia; excellentia → excellεntia; • the truncation of the personal pronoun io is regular in Ar1/Ar2 when needed due to metric reasons: io → i; • some plural endings are different: legna → legne; • in La the division of words is constant, whereas in Ar1/Ar2 the tendency is towards the synthetic writing: gia mai → giamai →etc.; • among the numerals, due is the one treated differently: La keeps the form of the ancient Italian (duo), whereas Ar1/Ar2 decline the form if singular (due) or plural (dui);18 • the verbal truncations of La are mostly eliminated in Ar1/Ar2: conoʃcer → conoʃcere; far → fare; par → pare etc.; • different treatment of the pronominal particle vi between La and Ar1/Ar2 when it introduces the complemento di termine: se → si; ve → vi.19 In addition to these formal variants that the systematic replacement of ε ω j ç and the distinction between u and v are all absent in La. From the point of view of substantives, the variant readings in the passage La→ Ar1/ Ar2 are the following:20 18. Arrighi’s humanistic training will have played a decisive role in the choice of the form. 19. Referred to here is the complement that specifies the receiver of an action. 20. Due to the Covid19 restrictions, it was not possible for me to carry out a complete collation of Ar1, but I could rely on the very recent study by Davoli 2020, 9, n. 18, in which the substantive variants in the passage La–Ar1 are reported. As for the collation La–Ar2, I was able to complete it personally, thanks to the digital reproductions of the manuscript and previous collations I made on some copies of the printing. In particular, the copies I examined are from Rome, Vatican Library, Membr.iv.18(int.2); Firenze, National Library, Palatino 2.9.3.37./a; Siena, Biblioteca degli Intronati, A xv C 019; Venezia, Biblioteca Marciana, Dramm. 0063. 001; Vicenza, Biblioteca Civica Bertoliana, Gonz. 006 002 014. 118 | Textual Cultures 14.1 (2021)

The manuscript classified as Additional 26873 (Hereafter La) is housed at the British Library in London and described in the online catalogue as follows: "Ital. Vellum; xvith cent. Injured by fire. Octavo. Giovanni Giorgio Trissino: Sophonisba: a tragedy". A few details may be added. La is part of a larger collection comprising Additional 26789-26876: eighty-eight items composed in various languages (English, French, Italian, Latin, Portuguese, Spanish), many of which have been damaged to different degrees by fire and damp. The scribe has been identified as Ludovico degli Arrighi (1475-1527), the well-known calligrapher from Vicenza (Veneto, northern-east Italy) who later turned into a typographer, active between his hometown and Rome during the first quarter of the sixteenth century. 1 Arrighi's renowned handwriting, defined as corsivo lodoviciano (Ludovico's italic), is a calligraphic italic based on the cancelleresca, well distinguished from the latter by the use of upright capitals characterized by long ascenders/descenders with curved ends, the lack of ties between characters, the greater space between transcriptional lines, and capital letters slightly higher than those of the current cancelleresca, whereas the inclination of the letters is the same, placed between 8 and 10 degrees (see Fig. 1). 2 La was not originally chartulated, nor paginated; a modern chartulation has been added in pencil in the upper right head of each charta (chartae 1-60); the sheets of each gathering are signed up until charta 4 with Latin letters together with Arabic numerals in sequence (see Figs. 2-3).
2. On Arrighi's italics see Casamassima 1963. For a general overview related to his printings see Gaskell 1972, 20-24.   The format is 8 o : A-G; each charta is roughly 1900 × 1300 mm, made up of nineteen poetic verses. Both the initial and final sections (chartae 1-17; 46-56) are those most damaged by fire (approximately 40-60% of the text is missing); the central section is sooty at the head of the chartae and some scraps are missing along the back, though generally speaking this section is fairly legible (see Fig. 4). 3 Considering the total absence of corrections and variant readings, La is a fair copy, although fragmentary. A note on the tail of the guard, added when the manuscript came to the British Museum, states that it was "presented by J. T. Payne on 29 July 1759". The news of a manuscript copy of the Sophonisba would not in itself be extraordinary since Trissino supervised the publication of three printed editions of the tragedy over five years: 1524 in July and again in September, both printed by Arrighi, and a third in Vicenza in 1529, printed by Ianiculo. 4 However, as Arrighi was the printer of the first edition, it is easy to insist on the importance of the information derived from such a source: like an autograph for literary authors, it tells us much about Arrighi's way of working, both as scribe and printer, and given the physical features of La, it follows logically to ask why Arrighi produced a copy of the tragedy and whether it is genetically related to Trissino's supervised editions. Preliminarily, it is possible to hypothesize the following: 1. La is a printer's copy for Arrighi's first and/or second edition of the Sophonisba (henceforth labeled, respectively, Ar1 and Ar2); 2. La is a printer's copy for an edition prior to the spelling reform of Trissino; 3. La is a copy composed by Arrighi and conceived for a restricted circulation within the Roman court of Pope Leone X (1475-1521).
Let us consider the first hypothesis. Given that La is calligraphic, free of variant readings, and composed of signed gatherings, it seems at first glance to be a printer's copy. At the same time, however, it lacks all of the typical indications of copy-preparation, such as annotations concerning 3. Due to the poor condition of the manuscript, it was not possible for me to carry out a typological analysis of the paper. I limit myself here, then, to giving the approximate height and width of the chartae. For the same reasons, nothing can be said about the original binding, since the manuscript was spun off and then reassembled following its restoration after the fire. 4. On the editions of the Sophonisba, see Davoli 2020. layout, italicization, capitalization, ending marks of the type-pages, or even traces similar to freshly made proofs, inky thumb-marks, etc. 5 Another point concerns the format: in the case of plays, format was often decided by convention, and it was quite common that the choice fell on the quarto book, though less expensive printed editions sometimes appeared in the octavo format. Both were handy formats, but the quarto was preferred, especially in the first quarter of the sixteenth century when the printed book had not yet become an entirely commercial product to be sold at low cost and when printers still aimed to provide their public with a quality product. In the case of Arrighi, a preference for the quarto can be assumed, considering that the outcome of his typography can be compared to that of a scriptorium in the quality of the paper, the page layout, and the design of the types. An accurate prevision of the length of the book was the first step in the preparation and organization of the edition, so as to come to a decision about the format and the right quantity of paper to be ordered; to this end, Arrighi or the compositor would have to cast off the manuscript copy by counting words according to the sizes of type and page that had been decided on. 6 The reasons to produce an accurate copy for the typography laid in the prediction of the exact contents of each type page: with the copy casted off, setting could begin anywhere in the book, avoiding the constraint, if necessary, of following the proper page order. Hence, since Ar1 and Ar2 are in quarto with twenty-two poetic verses per charta, we can hardly suppose that Arrighi produced as his copy an octavo manuscript with nineteen poetic verses per charta, for the difficulties that would have emerged in planning the printing.
The first hypothesis may also be tested by shifting the focus to the text. The comparison of La and Ar1/Ar2 reveals some important differences, the most outstanding of which concerns the failure to use Trissino's spelling reforms. Indeed, in a historical moment in which the Italian language was experiencing a phase of important codification and standardization, Trissino proposed a reform of the Latin alphabet that aimed to implement the writing system, especially in the representation of the open/closed traits of the vowels within literary Tuscan; this is because, particularly to non-5. Some authors gave clear instructions for the layout of their books, and give the great care that Arrighi reserved for the graphical aspects of his publications (both manuscript and printed), it is very likely that the fair copies he prepared included such indications, unlike La. See Greg 1966, 99-100. 6. There is no information on Arrighi's typography to indicate whether he was himself the compositor or just the master/overseer. See Smith 1755, 273 andMoxon 1962, 192. Tuscan speakers (the largest part of the inhabitants of the peninsula), how vowels were to be pronounced in certain phonetic contexts was far from clear due to the influence exerted by the respective dialects. Briefly, its reform included the following five additions to the current Ɛ and ω were the two additions that made the greatest impact on contemporaries: in this way each vowel within the Italian phonetic system had a corresponding grapheme. 8 What is relevant is that the Greek graphemes were used in opposition to the Latin ones: Greek for the open vowel, Latin for the closed one. 9 Again, given this context, that Arrighi wrote his copy 7. Given the focus of the present essay on the material and literary analysis of the manuscript, it would be inappropriate to address the complexity and vastness of the linguistic debate of the period. On the Questione della lingua (The debate around the Italian language), there is an almost endless bibliography; see, especially, the account in Tavoni [1987][1988][1989][1990]. Particular attention to the intertwining of literary and linguistic questions is given in Pozzi 1978Pozzi , 1988Pozzi , 1989 ; even among educated people the dialect or the italiano regionale -that is to say, the Italian spoken with significant differences in terms of phonetics, prosody and syntax among the macro-regions of Italy (northern, central and southern) -influences the way in which the open/closed traits of the vowel are employed according to its phonetic context, causing many doubts especially in formal contexts (i.e., work, public events, etc.). For the same reason, it is quite easy to understand from which region of Italy one comes from, and for the better linguists, perhaps specialized in single dialects/regions, it is also possible to distinguish between different areas/cities of origin. 9. In 1529 Trissino made changes to his spelling proposal due to criticisms raised by Tuscan literates, and, in particular, by Claudio Tolomei ( without using Trissino's additions is hardly sustainable due to the difficulties that would have emerged during the composition as well as proof-correction phases: even the ablest compositors and correctors -among whom certainly Arrighi or someone working for him -made mistakes, and the fact that the new Greek letters had to be introduced with care cannot have led Arrighi to produce a copy for the typographer without employing them. Moreover, Trissino was at the time in Rome and very likely supervised the proofreading, at least in the case of Ar2, since the correction phase was seldom left to the compositor alone. In the case of Arrighi's Sophonisba, it is also possible that he employed a professional corrector and not, as was more usual, a piece-worker who had to pay attention to his own mistakes in his own time, and who would have been tempted to overlook such errors as they would have taken longer to mend. 10 It is no coincidence that, following the release of Ar1, which generated a heated controversy among the detractors of his integrated alphabet, Trissino prevented further criticism by making a reprint (Ar2) of the tragedy only two months later. Discarding thus the first hypothesis, La could instead be a manuscript conceived by Arrighi for an edition prior to Trissino's spelling reform. In fact, what makes us inclined towards the hypothesis of a printer's copy is the presence of signatures on the sheets. As is well-known, the use of signatures in printing derives from the assembly of sheets in order to get them correctly oriented and in the right order. For the same reason, as the format decreases the number of chartae within the gathering increases, so that signatures were also placed on the rectos of a few leaves after the first of each sheet in order to help the binder in folding. Their presence is therefore an important fact and must be properly analyzed to draw the necessary conclusions. The signatures, both in La and Ar1/Ar2, are made up of letters from the Latin alphabet, a convention deriving from manuscript culture; manuscript culture also supplies the mechanism of the twenty-fourth folded sheet signed with the double letter (for instance ee, ff etc.), as is the a substantial modification of the previous alphabet reverses the use of ω, later employed for [o], while ε remains unaltered. This last change was introduced by Trissino according to his knowledge of the Greek language (ω corresponding to [ɔ]); but this was at the expense of the symmetry of the spelling system, which appeared to his detractors even less functional and coherent than the previous one. 10. Mistakes in individual letters always occurred in some numbers no matter how skilled the compositor; such mistakes were inevitably compounded in cases where it was necessary to add Greek letters to the Latin alphabet; see Gaskell 1972, 110-16. case of La (charta 41r), whereas Ar1/Ar2 do not maintain this technical usage (see Fig. 5).
Another difference among La, Ar1, and Ar2 is that the manuscript has signatures until the fourth sheet, distinguished by letters together with Arabic numerals, whereas letters and Roman suffixes (e.g., aij, bij) are only used in Ar1/Ar2 up to the second sheet. 11 Moreover, even though La is fragmentary, charta 4r is signed a4: going backwards, charta 1r coincides with sheet a1, so that we suppose the signing of preliminary leaves was made by symbols rather than letters for the impossibility of returning the dedication to the counting; 12 on the other hand, in Ar1/Ar2 the signing begins with the title-page, as the sheet introducing the dedication is signed aij (the structure is also confirmed by the register). Headlines with running titles are absent from La and Ar1/Ar2, as well as catchwords at the end of the direction line and chartulation/ pagination; in the case of Ar1/Ar2, the lack of these practices, intended to help the compositor set the pages in the right order for the printing, would have made their arrangement in the proper order very slow, so that the print run could not have been very high. Therefore, it is very likely that the edition was produced for a limited audience. This tells us a lot about the way in which Arrighi interpreted his work as scribe and typographer. Because he was first a humanist and a calligrapher, he intended the printed book as a valuable product and, consequently, he created a 11. This is due to the smaller format of La, for which it was necessary to mark more than two sheets, since each gathering was made up of eight chartae, and therefore would have created greater difficulties during the binding phase. 12. La, charta 1r, coincides with charta c3r of Ar1/Ar2, so that almost three gatherings are not coinciding and must have been signed using symbols rather than the Latin alphabet. sort of hybrid object, strongly influenced by the manuscript culture and at the same time open to the practices of the new medium. For that reason, his manuscripts must be treated cautiously, even though they could appear on first glance as printers' copies. The truth is that Arrighi's printed editions distinguished themselves from most of his contemporaries' books in that Arrighi's resembled more a literal reproduction of the manuscript rather than a normalized version of it with substantial changes designed to optimize the typographic work. In this respect, La represents the opposite situation: very similar to a printer's copy, it nonetheless lacks those necessary elements which would have qualified it as such -instructions about the layout, italicization, capitalization, etc. -making it most likely to be a handwritten copy of the Sophonisba written before 1515. Indeed, we know that Trissino composed his tragedy between 1513 and 1515. During this period Morsolin (1894) reports the popularity of some of its scenes in Rome even before the premiere. 13 In support of this historical data, in two letters Giovanni Salviati (1490-1553 claims that manuscript copies of the tragedy, profoundly incorrect, circulated in large numbers in the city at the time, right after Trissino finished writing it. 14 The argument can be taken to further support the fact that La is antecedent to 1518 -the dating proposed by Mazzoleni -assuming it was composed precisely during this period of great fortune for the tragedy. In addition, Arrighi was already in Rome in that period and had not yet established his typographical activity (Pratesi 1962). In light of this historical and biographical data, La is more likely to be a handwritten copy produced during the Roman period in the lively circulation of the tragedy rather than a printer's copy for an edition prior to Ar1: thus for all intents and purposes La is a scribal publication.
13. For all it is known, the Sophonisba was staged publicly for the first time only many years later in Vicenza, in the great hall of the Basilica (1562), with a stage set conceived by the architect Andrea Palladio (1508-1580), pupil of Trissino, and painted by Giovanni Antonio Fasolo (1530-1572); see Magrini 1847. 14. Giovanni Salviati, while complaining to Trissino for not sending a copy of the tragedy, however reassured him that he was aware of the difficulties of finding a scribe who would have copyed it "a suo modo" (in Trissino's own way), which seems to be an allusion to the use of the new letters. Nonetheless, Salviati claims to have found some manuscript copies and, although they were very incorrect, had them transcribed. Salviati traced back the errors to the fact that the tragedy was copied within one night; despite that, he affirmed that many other copies were transcribed. See Trissino 1729, xvi; Salviati 1878; Morsolin 1894, 77. In addition to this third hypothesis, a complete collation among La, Ar1, and Ar2 shows that the first witnesses an antecedent stage of the draft, with a relevant number of accidentals and substantive variants. Besides, unlike what Salviati stated regarding the flawed nature of the copies he saw during his Roman stay, La is definitely correct, reflecting the fact that Arrighi worked very accurately, both as a copyist and as a typographer. Here below is a summary of the systematic variants in the passage from La to Ar1/Ar2: 15 • in La the form of the coordinating conjunction is et, whereas Ar1/ Ar2 use alternately ε/εt depending if it is followed by a word beginning with a vowel (εt) or a consonant (ε); • in La the articulated prepositions are written in analytical form, while the synthetic one is used in Ar1/Ar2 (a 'l → al; de 'l → del etc.); • as regards punctuation, a systematic intervention appears at the end of the poetic verses: where in La there is a full stop or two points, Ar1/Ar2 substitute them with the semicolon; further, in La the circumflex accent is used to point out the stressed syllable of the past tense (passato remoto), whereas in Ar1/Ar2 is the acute accent (andô → andὼ; formô → formὼ etc.); occasionally, La accentuates completely independently from Ar1/Ar2, such as in charta 27v Ar2, the trend towards a widespread elision is more marked, as reported; see Davoli 2020, 17. 17. In those years diphthongization was in the literary language assuming its regularized form but was still not predominant.
• La is less consistent in pseudo-etymological spellings if compared to Ar1/Ar2: ancora/ancor → anchora/anchor; qualcuno/qualcun → qualchuno/qualchun; some are retained or restored in La: chara → cara; loco → luωgo; palagio → palazo; • La is inconsistent in the treatment of the Latin ending -TIA: patienza → patiεntia; reʃiʃtenza → reʃiʃtεntia; temperanza → temperantia; but also: ʃententia → ʃentεntia; excellentia → excellεntia; • the truncation of the personal pronoun io is regular in Ar1/Ar2 when needed due to metric reasons: io → i; • some plural endings are different: legna → legne; • in La the division of words is constant, whereas in Ar1/Ar2 the tendency is towards the synthetic writing: gia mai → giamai →etc.; • among the numerals, due is the one treated differently: La keeps the form of the ancient Italian (duo), whereas Ar1/Ar2 decline the form if singular (due) or plural (dui); 18 • the verbal truncations of La are mostly eliminated in Ar1/Ar2: conoʃcer → conoʃcere; far → fare; par → pare etc.; • different treatment of the pronominal particle vi between La and Ar1/Ar2 when it introduces the complemento di termine: se → si; ve → vi. 19 In addition to these formal variants that the systematic replacement of ε ω j ç and the distinction between u and v are all absent in La. From the point of view of substantives, the variant readings in the passage La→ Ar1/ Ar2 are the following: 20 18. Arrighi's humanistic training will have played a decisive role in the choice of the form. 19. Referred to here is the complement that specifies the receiver of an action. 20. Due to the Covid19 restrictions, it was not possible for me to carry out a complete collation of Ar1, but I could rely on the very recent study by Davoli 2020, 9, n. 18, in which the substantive variants in the passage La-Ar1 are reported. As for the collation La-Ar2, I was able to complete it personally, thanks to the digital reproductions of the manuscript and previous collations I made on some copies of the printing. In particular, the copies I examined are from Rome, Note that Morsolin (1894, 463-64) briefly examines the comparison between Ar1/Ar2 and points out that the first, charta i3r, contains 23 poetic verses instead of the 22 found in the latter, thus diverging the count of the verses starting from there. Furthermore, on c. m4v, 6-7, the reiteration of the interjection Hωimεi is reported as an error, which Davoli (2020, 13) justifies as an optical jump during the composition, for the iteration of the word a few lines below that would have generated the error. However, the repetition proceeds with ascending climax, since Hεrmina pronounces first one, then two and finally three times Hωimεi in sequence (m4v, vv. 7, 10, 11, 15, 16, 17): hence, it is not an error (See Fig. 6).
The conclusions that can be drawn from the analysis of the variant readings between La and Ar1/Ar2 highlight the cultural intent and the poetic practice of Trissino, supporting the hypothesis that, given the lack of all the linguistic features that will later characterize the prints, La was copied by Arrighi shortly after the tragedy was ended by the author. Beyond the linguistic variants, mainly formal, the substantive ones are coherent with the changes in the expressive form of the author, which particularly concern the syntactic structure of the periods and the choice of lexemes, since the expressive modality of the tragedy had to be, according to Aristotle's Poetics, as close as possible to the common language (sermo humilis). 21 Indeed, the direction of corrections such as those on sheet d1r shows the tendency to a greater linearity of verbal speech, characterized by paratactic syntax with frequent enjambements in favor of the continuity of the poetic speech, in this supported by the blank verse and by the alternation of hendecasyllables and seven-syllables verses, with the clear preponderance of the latter. For all these reasons, La is configured as a dedication copy that Arrighi probably composed and presented to Leone X; as a result, its dating should go back to at least 1515, and in all probability, 1514.
Finally, it is worth dedicating one last note to the cultural context in which La was produced, especially concerning the revolution of the media occurring in Italy in the early sixteenth century. Indeed, although printing was no longer a novelty, nevertheless in the first quarter of the century it was still at an experimental stage, and the great interpreters of the new medium, including Manutius and Arrighi, were first of all humanists fully trained in the culture of manuscript books. This, as is particularly evident in the case of Arrighi, had greatly influenced these printers' ways 21. These theoretical and poetic aspects of Trissino are well illustrated in the I and V divisions of his Poεtica (Vicenza, Ianiculo 1529). A modern edition is in Weinberg 1970. of understanding the typographical product, to such an extent that both Arrighi's and Aldo Manuzio's printed editions may be defined as hybrid objects, influenced to a large extent by the culture of the manuscript. Yet, if this is true in the sense of an influence of the culture of the manuscript book on that of the printed book, it must not be forgotten that the opposite influence was possible, and actually happened: the new printing medium, which at that time was still developing working practices while consolidating the existing ones, had a strong influence chiefly on the layout of manuscript products, to such an extent that some artifacts may seem, at first glance, printer's copies. Indeed, they must not be de-contextualized by other determining elements for identification, such as those already mentioned (annotations, typical signs of the typographical work on the copy, etc.). Precisely for this reason, La remains a very valuable witness that allows us to evaluate a previous draft of the Sophonisba, determining the variation in the passage La → Ar1-Ar2. At the same time La returns to us the figure of Arrighi as a copyist, an example of the level of professionalism at which such a competent and talented scribe/typographer could reach.