Gazes ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 ISSN 2159-3159 Article copyright held by the author(s) and made available under the Creative Commons Attribution License CC-BY https://creativecommons.org/licenses/by/4.0/ The Lost Poems on the Muses by Theodoros Gazes Julián Bértola and Anthony Ellis HEODOROS GAZES (ca. 1400/10–1475/6) is best known as a teacher and translator of Greek in Renaissance Italy.1 From two reports by his former student, Ludovico Carbone, we knew that he also composed poems on the Muses, inspired by a series of pictures in Leonello d’Este’s studiolo at the palace of Belfiore in Ferrara.2 Until recently this poetic cycle was thought to be lost, with the partial exception of two poems pre- served in Latin translation by Carbone. But the full cycle of epigrams has now come to light in a late-15th-century copy of Herodotus’ Histories made by Demetrios Raoul Kabakes (Vat.gr. 1359). Kabakes put the poems to a new purpose: as epigrams to decorate each of Herodotus’ nine books—traditionally named after the Muses. These poems were recently made available 1 For an overview of Gazes’ life and career see PLP no. 3450; D. J. Geanakoplos, “Theodoro Gaza, a Byzantine Scholar of the Palaeologan ‘Renaissance’,” in Constantinople and the West (Madison 1989 [1984]) 68–90; C. Bianca, “Gaza, Teodoro,” in Dizionario Biografico degli Italiani LII (Rome 1999) 737–746. 2 The studiolo was eventually destroyed and the paintings dispersed to several collections. On the now lost studiolo and the reconstruction of the pictorial cycle of the Muses see A. K. Eörsi, “Lo studiolo di Leonello d’Este e il programma di Guarino da Verona,” Acta Hist.Art.Hung. 21 (1975) 15–52; A. Mottola Molfino et al. (eds.), Le Muse e il principe: Arte di corte nel Rinascimento padano (Modena 1991); A. K. Eörsi, “Da Medea attraverso l’Amore a Ter- sìcore: Nuovi appunti alle rappresentazioni delle Muse nello studiolo della Villa Belfiore,” Acta Hist.Art.Hung. 45 (2004) 3–23; G. Manni, Belfiore: Lo studiolo intarsiato di Leonello d’Este (Modena 2006); and the online resources available at https://studiolobelfiore.unibo.it. T JULIÁN BÉRTOLA AND ANTHONY ELLIS 481 ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 through the Database of Byzantine Book Epigrams (DBBE).3 Since the autograph copy has been lost and the only witness is in the hand of an often unreliable scribe, this article offers a critical edition and translation of one of the few known examples of poetry by Gazes.4 The edition is preceded by an introduction, reviewing the circumstances in which the poems were composed, and an overview of their transmission and what it reveals about the contacts within the Greek diaspora in Italy after the Ottoman conquests. 1. The occasion for the composition of the epigrams After a period in Mantua, Theodoros Gazes reached Ferrara in 1446 and remained there until 1449, teaching Greek litera- 3 https://www.dbbe.ugent.be; see DBBE Occurrences 27063, 27066, 27067, 27068, 27070, 27072, 27074, 27076, and 27077 (created by Bértola). The epigrams were transcribed from Vat.gr. 1359 by Girolamo Amati in the handwritten catalogue of the beginning of the 19th century (Inventarium codicum Vaticanorum Graecorum 993–2160) preserved in the Biblioteca Apostolica Vaticana, Sala Cons. MSS. 323, ff. 355v–356r (available online at https:// digi.vatlib.it/view/INV_Sala.cons.mss.323.rosso/0358). 4 Besides the cycle on the Muses, four other book epigrams can be attrib- uted to Gazes with more or less certainty: two on Homer in hexameters in a manuscript commissioned by Francesco Filelfo (Laur.Plut. 32.1, ff. 17r, 654v; see DBBE Types 6586 and 3479); another on Homer in elegiacs in a manu- script which belonged to Bessarion (DBBE Type 5288); another in iambic trimeters in a manuscript which belonged to Filelfo (Vat.gr. 1334, f. 107v; see DBBE Type 6035). On Gazes as a scribe and on his books see D. Speranzi, “ ‘De’ libri che furono di Teodoro’: una mano, due pratiche e una biblioteca scomparsa,” Medioevo e Rinascimento 23 (2012) 319–354. There are also two poems in elegiacs dedicated to Ciriaco of Ancona: one edited by A. Olivieri degli Abati, Commentariorum Cyriaci Anconitani nova fragmenta notis illustrata (Pesaro 1763) 37, which is also preserved in MS. Rome, Biblioteca Vallicel- liana, Allacci 142, f. 231r, and another poem first edited by D. A. Zakythinos, “Poèmes inédits de Ciriaco d’Ancona (avec une épigramme de Théodore Gaza),” ByzZeit 28 (1928) 270–272, and, more recently, by F. di Benedetto, “Fetonte e i pioppi (e la zanzara) in un epigramma di Teodoro Gaza,” in S. Bianchetti et al. (eds.), ΠΟΙΚΙΛΜΑ. Studi in onore di Michele R. Cataudella (La Spezia 2001) 383–396. Fillippomaria Pontani is working on a new edition of the latter two poems, and the book epigram for Bessarion. 482 POEMS ON THE MUSES BY THEODOROS GAZES ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 ture and studying medicine.5 He made a lasting impression on his students. One of them, Ludovico Carbone, repeatedly cele- brated his teacher in his orations and dialogues. In his Dialogus de amoenitate, utilitate, magnificentia Herculei Barchi (composed in 1475/6), a eulogy of the hunting reserve designed by Ercole d’Este at Belfiore,6 Carbone refers to the studiolo of the Muses begun by Leonello and finished by his successor, Borso, who was in turn Ercole’s predecessor. In a short excursus on painting and poetry, Carbone first mentions Angelo Maccagnino of Siena (described as the painter of two of the Muses) and Cosmè Tura (who painted the other portraits), and recalls four lines of Greek poetry composed impromptu by Gazes at Leonello’s request, on the completion of the paintings.7 While Carbone and his inter- locutor roam the studiolo, the latter asks for a translation of the verses into Latin, because he does not know Greek. Carbone replies:8 5 On Gazes’ stay in Ferrara see J. Monfasani, “L’insegnamento di Teodoro Gaza a Ferrara,” in Greeks and Latins in Renaissance Italy: Studies on Humanism and Philosophy in the 15th Century (Aldershot 2004 [1994]) III. 6 The dialogue was edited by A. Lazzari, “Il ‘Barco’ di Ludovico Carbone,” Atti e Memorie della Deputazione Ferrarese di Storia Patria 24 (1919) 4–44; an Italian translation of the relevant passage can be found in A. Di Lorenzo, “Le Muse di Belfiore nelle descrizioni degli umanisti,” in Le Muse e il principe 321–331, at 328–331. 7 Lazzari, Atti e Memorie 24 (1919) 34–35: “Agnosco duas ex his picturis ab Angelo Senensi artificiosissime laboratas … Caeterae a Cosmo nostro, pic- tore nobilissimo, non minori arte perfectae … Nunc mihi in mentem veniunt quattuor versiculi graeci quos Theodorus ille magister meus, vir omnium qui vivant et doctissimus et humanissimus, a Leonello rogatus de his Musis, quae tunc absolutae erant, ex tempore edidit” (“I recognize two of these paintings made very artfully by Angelo of Siena … The rest have been completed by our Cosmè, most noble painter, with no less art … Now four Greek verses come to my mind, which that teacher of mine, Theodoros, one of the most learned and cultivated men alive, as requested by Leonello, spontaneously produced on these Muses, which had been finished at that time”). Note that Carbone refers to Gazes as still living. All translations in this paper are ours, unless otherwise indicated. 8 Lazzari, Atti e Memorie 24 (1919) 35. JULIÁN BÉRTOLA AND ANTHONY ELLIS 483 ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 Animadverte quantum antiquitatem redolent. “Nomen mihi est Clio, quoniam virorum laudes canto. Auream vestem fero, et aureum animum habeo. Melpomene sum, o hospes, puella Iovis aeterni; cantu delectans mortales, aspectu vero immortales.” Note how they breathe antiquity! “My name is Clio, because I sing the praises of men. I wear golden attire, and have a golden soul.” “Stranger, I am Melpomene, daughter of eternal Zeus, who delights mortals with her song and immortals with her look.” Later, the interlocutor asks Carbone to versify his Latin trans- lation. Carbone agrees and offers a display of his poetic skills:9 Clio mihi nomen celebranti gesta virorum, Aurea mî vestis, aureus est animus; Melpomene dicor superni filia Regis, Voce trahens homines luminibusque deos. Clio is my name, as I celebrate the deeds of men, golden is my attire, golden my soul; Melpomene I am called, daughter of the celestial King, who attracts men with her voice and the gods with her eyes. As the reader will notice, the prose translation of Carbone faith- fully reproduces the Greek of poems 1 and 4 from the cycle published below. The poetic version is rather freer, though in a verse form which reproduces the elegiacs of Gazes’ original.10 The identification of these verses with Gazes’ Greek distichs in the MS. of Herodotus (Vat.gr. 1359) is conclusive. But we can delve deeper into the context in which the poems were produced and circulated. A preliminary question about Carbone’s report is why he cites only two poems from a series of nine. This may be connected with his claim to be reciting the verses from mem- ory (“Nunc mihi in mentem veniunt...”), although the fictitious 9 Lazzari, Atti e Memorie 24 (1919) 37. 10 Before the poems of Gazes came to light, Nigel Wilson had attempted to reconstruct the original Greek from Carbone’s Latin version, with a number of successful suggestions: οὔνοµ’ ἐµοὶ Κλειώ, ὅτι ἀείδω κλέα ἀνδρῶν. / εἵµατα χρύσε’ ἐγὼ καὶ διάνοιαν ἔχω. / Μελποµένη ’γώ, ξεῖνε, Διὸς παῖς αἰὲν ἐόντος, / τέρπω ὀπὶ θνητούς, σχήµατι δ’ ἀθανάτους. See N. G. Wilson, “Guarino, Gio- vanni Tzetze e Teodoro Gaza,” in Le Muse e il principe 83–86, and “Greek Inscriptions on Renaissance Paintings,” IMU 35 (1992) 215–252, at 221. 484 POEMS ON THE MUSES BY THEODOROS GAZES ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 setting of the dialogue should discourage us from accepting the scenario at face value. But his apparently spontaneous recollec- tion of how the poems came to be composed seems to contain important details about the first performance and later transmis- sion of these poems. Poetic improvisation (“ex tempore edidit”) could explain the paucity of manuscript evidence of Gazes’ poems. A fuller picture emerges from another text by Carbone. In an oration on the liberal arts, he implores Leonello d’Este to have Gazes’ verses inscribed in his studiolo. After declaring them comparable in style to the works of Callimachus, Propertius, and Tibullus, Carbone says that Gazes’ poems name the Muses, de- scribe their various offices, and allude to the paintings.11 Since 11 The passage is quoted in Lilio Gregorio Giraldi’s Dialogi duo de poetis nostrorum temporum, first published in 1551, now edited with English translation by J. N. Grant, Lilio Gregorio Giraldi: Modern Poets (Cambridge 2011) 112–113: “Sed de Gaza quaeso audite quae Ludovicus Carbo in quadam de artibus liberalibus oratione ad Leonellum Estensem Ferrariae principem scribit, in qua inter ceteras Gazae praeceptoris laudes et hoc ait: ‘Dicam equidem quanto dignus Theodorus honore, quem plus quam dimidium animae meae semper diligam? Cuius versiculos, quos de Musis tuis elegantissimos edidit, in sacrario illo pulcherrimo incidi iubeas etiam atque etiam rogo et obsecro. Quid enim gravius Callimachus, quid suavius Propertius, quid pulchrius Tibullus dicere umquam potuisset? Nam et nomen et officium Musarum con- cinne exprimunt et earum laudem ad picturam alludentes brevius concinunt; nobile hominis ingenium tantum philosophum in huius modi deliciis prae- stare’ ” (“But please listen to what Ludovico Carbone writes about Gaza in an oration about the liberal arts, dedicated to Leonello d’Este, duke of Fer- rara. Among all the plaudits he gives to his teacher Gaza he says this too: ‘What great honor shall I say Theodore deserves, a man whom I shall always love more than life itself? I ask and plead with you again and again that you give the order for Theodore’s very elegant verses on your Muses that were produced by him to be engraved in that most beautiful shrine. Could Cal- limachus have written anything more powerful, Propertius anything more sweet, Tibullus anything more beautiful? For the verses most pleasingly ex- press the name and function of the Muses and express harmoniously and briefly his praise of them in their allusion to the painting. The noble genius of the man in composing such delightful verses surpasses his greatness as a philosopher’ ” [transl. Grant, slightly modified]). See also the editions by JULIÁN BÉRTOLA AND ANTHONY ELLIS 485 ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 the oration is addressed to Leonello, it is presumably to be dated to some point before his death in October 1450.12 By that time, Carbone implies, Gazes had composed poems, taking the paint- ings as his inspiration. Nevertheless, his works seem not to have found a place in the studiolo. Confirmation comes from the merchant and antiquarian Ciriaco of Ancona. In a letter dated July 1449 we learn that one of the paintings bore an epigram, though not one from the cycle composed by Gazes. Ciriaco first describes a visit to Ferrara in which Leonello showed him a painting of Christ’s Deposition by Rogier van der Weyden. Then, at the palace in Belfiore, he saw the two Muses whose portraits had been completed, Clio and Melpomene, painted by Angelo Maccagnino.13 Ciriaco’s de- scription of these two Muses closely matches the traits and K. Wotke, Lilius Gregorius Gyraldus: De poetis nostrorum temporum (Berlin 1894), and by C. Pandolfi, Lilio Gregorio Giraldi da Ferrara: Due dialoghi sui poeti dei nostri tempi (Ferrara 1999), with Italian translation; the brief biographical essay on Theodoros Gazes by Allacci, reproduced in PG 161.973–974; and S. J. Campbell, “ ‘Sic in Amore Furens’: Painting as Poetic Theory in the Early Renaissance,” I Tatti Studies in the Italian Renaissance 6 (1995) 145–168, at 154– 155. 12 In fact, this oration could correspond to De litteris et studiis, dated to 1448; see L. Paoletti, “Carbone, Ludovico,” in Dizionario Biografico degli Italiani XIX (Rome 1976) 699–703, at 699. 13 Latin text and English translation from E. D. Bodnar, Cyriac of Ancona: Later Travels (Cambridge 2003) 366–367: “Cuius nempe inclytae artis et exi- mii artificum ingenii egregium equidem imitatorem, Angelum Parrisium, quoque Senensem, recens picturae in Latio specimen vidimus … At enim vero Clio Melpomeneque, quarum iam absolvisse figuras vidimus…” (“To be sure, we did see in Latium a recent example of [this kind of] painting, [by?] a remarkable imitator of famous art and of the extraordinary genius of artists, Angelo from Ferrara(?), also from Siena … But as for Clio and Melpomene, whose figures, we saw, had already been completed…”). For other transla- tions of this passage see M. Baxandall, “Guarino, Pisanello and Manuel Chrysoloras,” JWarb 28 (1965) 183–204, at 187–188, and Di Lorenzo, in Le Muse e il principe 326–327. Note that the ascription of two Muses to Angelo Maccagnino, like the exclusive mention of Clio and Melpomene, square with the report of Carbone’s Dialogus de amoenitate. 486 POEMS ON THE MUSES BY THEODOROS GAZES ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 actions picked out by Gazes in the epigrams edited below. Clio, for instance, is attired in gold, holds a trumpet and an open book, and “appears to inspire men to glory” (“homines ad gloriam excitare perspicitur”). Melpomene, in turn, is also sumptuously dressed, plays a cithara and “seems to shape her voice with rose- red lips” (“roseis labiis vocem formare visa”), as if she were singing.14 Ciriaco closes his ekphrasis by stating:15 “Clio has this inscription, composed by our Guarino, written on the base: Through histories, I preserve both fame and ancient deeds.” This Latin hexameter is also attested in a famous letter from Guarino of Verona to Leonello d’Este, dated to November 1447, almost two years before Ciriaco’s account.16 Guarino commends Leonello for his resolution to have the Muses depicted (“in pingendis musis”).17 He then succinctly describes the attributes of each of the nine Muses (although he concedes that the number is contested), which he views as symbolic of various aspects of human inventiveness and intellectual endeavor. Guarino’s letter simultaneously instructs Leonello on how each Muse should be represented. Two poems in hexameters are appended to the letter in some manuscripts. The first poem is, in fact, a collection of nine monostichs, and its first verse corresponds to the inscrip- tion reported by Ciriaco:18 14 Text and translation Bodnar, Cyriac 366–369. 15 Bodnar, Cyriac 368–369: “Clio hoc ad basim ex Guarino nostro epi- gramma conscriptum habet: Historiis famamque et facta vetusta reservo.” 16 R. Sabbadini, Epistolario di Guarino Veronese II (Venice 1916) 498–500, no. 808, with commentary in III (Venice 1919) 406. The Latin is reproduced with a facing Italian translation in Di Lorenzo, in Le Muse e il principe 322–325. See also Baxandall, JWarb 28 (1965) 186–189, 201–202; Eörsi, Acta Hist.Art.Hung. 21 (1975) 21–27, 30–43. On the poems attached to the letter see A. Tissoni Benvenuti, “Lettera a Leonello d’Este sulle Muse (5 novembre 1447),” in Le Muse e il principe 158–161. 17 Sabbadini, Epistolario 498. 18 Sabbadini, Epistolario 500. JULIÁN BÉRTOLA AND ANTHONY ELLIS 487 ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 ΜΟΥΣΑΙ Clio. Historiis famamque et facta vetusta reservo. Thalia. Plantandi leges per me novere coloni. Erato. Connubia et rectos mortalibus addit amores. Euterpe. Tibia concentus hac praemonstrante figurat. Melpomene. Haec vivos cantus et dulcia carmina format. Terpsichore. Ista choris aptat saltus ad sacra deorum. Polymnia. Haec docuit segetes acuens mortalia corda. Urania. Signa poli, varias naturas monstro viasque. Calliope. Materiam vati et vocem concedo sonantem. Muses Clio. Through histories, I preserve both fame and ancient deeds. Thalia. From me farmers have learned the laws of planting. Erato. She gives marriages and fitting loves to mortals. Euterpe. While she conducts, the flute weaves harmonies. Melpomene. She composes lively songs and sweet poems. Terpsichore. She provides the dancers with the steps for the rites of gods. Polymnia. She taught of the crops, sharpening men’s wits. Urania. I show the constellations of the heavens, their various natures and ways. Calliope. I offer the poet a subject and a resonant voice. The verses summarize the allegorical descriptions which Guarino provides in the body of the letter as instructions for the pictorial cycle. Guarino’s guidelines show a clear correspon- dence to Ciriaco’s descriptions of the two Muses, and it seems likely that the letter and its poem influenced the preparation of the paintings. Despite mentioning the Latin epigram with the painting of Clio, Ciriaco makes no mention of any corresponding epigram for Melpomene.19 But a painting of the Muse Thalia, now in the Museum of Fine Arts of Budapest (inv. no. 44), preserves the 19 In comparison to Clio, the painting of Melpomene described by Ciriaco follows the instructions of Guarino less closely (Guarino, for instance, recom- mends that the Muse hold a book with musical notes, whereas Ciriaco says that she holds a cithara). However, both Ciriaco and Guarino focus on Mel- pomene as representation of singing and vocals. 488 POEMS ON THE MUSES BY THEODOROS GAZES ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 corresponding epigram from Guarino’s letter.20 The Muse is depicted in line with Guarino’s recommendations: Thalia, as her name indicates, invented the agricultural art of sowing, and is thus adorned with flowers, branches, ears of grain, grapes, and other fruits.21 The Muse is portrayed sitting on a bejeweled throne with her feet on a platform. Below the platform, the em- blem of Borso d’Este (who ruled after Leonello, 1450–1471) and two wooden panels are depicted at the bottom of the painting. Each of these panels contains an epigram in majuscules, de- signed to resemble an inscription. The one on the right displays Guarino’s hexameter: PLANTANDI LEGES PER ME NOVERE COLONI. The one on the left has a Greek hexameter, apparently a translation of Guarino’s Latin verse: ΦΥΤΕΙΑΣ ΓΕ ΝΟΜΟΥΣ ΑΠ ΕΜΟΥ ΔΗ ΓΝΩΤΕ ΓΕΩΡΓΟΙ. There is yet another text in this painting: a small piece of paper protrudes from the pedestal, as if hanging from the top of it. On the paper is written EX MICHAELE PANONIO, an indication of the painter: Michele 20 An image is available at the website of the Museum, https://www.mfab. hu/artworks/the-muse-thalia/, and at https://studiolobelfiore.unibo.it/index .php/talia/. On this painting see Eörsi, Acta Hist.Art.Hung. 21 (1975) 16, 23, and V. Tátrai, “La Musa Talia,” in Le Muse e il principe 404–408; on the epi- grams in particular see Wilson, IMU 35 (1992) 218–220. 21 Wilson, in Le Muse e il principe 83–84, maintains that Guarino’s interpreta- tion of some Muses as related to domains outside of literature depends on the 12th-century scholar Ioannes Tzetzes, although he admits more ancient antecedents. This conception of the Muses is not unparalleled in other Byzantine authors. For example, we can mention the polymetric cycle on the Muses by Euthymios Tornikes dedicated to Isaac II Angelos, edited by A. I. Papadopoulos-Kerameus, Noctes Petropolitanae (St. Petersburg 1913) 188–198; see N. Zagklas, “Metrical Polyeideia and Generic Innovation in the Τwelfth Century: The Multimetric Cycles of Occasional Poetry,” in A. Rhoby et al. (eds.), Middle and Late Byzantine Poetry: Texts and Contexts (Turnhout 2018) 44– 70, at 52–55. On Guarino’s instructions as embedded in Renaissance aesthetic polemics about poetry, art, and the decency of the classics, see Campbell, I Tatti Studies in the Italian Renaissance 6 (1995) 145–168, and “The Traffic in Muses: Painting and Poetry in Ferrara around 1450,” in L. Benedetti et al. (eds.), Gendered Contexts: New Perspectives in Italian Cultural Studies (New York 1996) 49–68. JULIÁN BÉRTOLA AND ANTHONY ELLIS 489 ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 Pannonio. The painting is commonly dated to 1456/7, the transition years after the death of Maccagnino and before Tura took over the pictorial enterprise. The Greek inscription on the left seems to have an implicit priority over its Latin equivalent on the right, due both to the directionality of Greek and Latin script and to the traditional view of Greek as the bedrock of Latin literature. But it seems more likely that the Greek hexameter is a later translation of the Latin line, since the latter is part of a poetic series which precedes the paintings themselves, attested in the appendix to Guarino’s letter.22 The Greek line has been tentatively attributed to Gazes himself,23 but does not correspond to the poem on Thalia edited below. The painted wooden panels may explain Carbone’s unusual choice of words when he besought Leonello to have Gazes’ verses inscribed.24 Moreover, the tablets give the impres- 22 The translation’s most notable departure from Guarino’s verse—pre- sumably the result of the translator’s zeal to observe the metre—is the direct address to the farmers in the second-person plural, which can be either an aorist imperative or an unaugmented indicative: “Farmers, learn from me the laws of planting” or “Farmers, from me you learned the laws of planting” (Φυτείας γε νόµους ἀπ’ ἐµοῦ δὴ γνῶτε γεωργοί). 23 Wilson, in Le Muse e il principe 84–85, and IMU 35 (1992) 219. 24 Wilson, in Le Muse e il principe 85–86, and IMU 35 (1992) 220–221, voices puzzlement that Carbone specified that the epigrams of Gazes should be “carved out” (“incidi”), as if the Muses were sculpted in the studiolo, and suggests that Carbone imagined an inscription on the wooden frames or on the wall adjacent to the pictures; Manni, Belfiore 104, suggests that Carbone was thinking of wood-inlaid inscriptions. Now, the Muses were certainly not sculpted, as Carbone says that the poems allude to paintings (“ad picturam alludentes”). An explanation may lie in the representation of the verses in the Budapest Thalia. The location of the inscribed tablets in the painting of Thalia also fits Ciriaco’s description (“Clio hoc ad basim ex Guarino nostro epigramma conscriptum habet”). At least two other paintings identified as part of the cycle from Belfiore show traces of similar panels painted into the portrait below the pedestal: Terpsichore, now in the Poldi Pezzoli Museum in Milan, inv. no. 1559 (https://museopoldipezzoli.it/scopri/collezioni/ catalogo/opera/?guid=8045de9e-43b6-437b-a222-abb087db54ca) and 490 POEMS ON THE MUSES BY THEODOROS GAZES ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 sion of being too big for the little text they contain, as if they might have been designed to accommodate a distich or even more verses. Be that as it may, by the end of 1450 Leonello has died and Gazes has left Ferrara. The project of completing the studiolo was taken over by Leonello’s successor, Borso d’Este, and the Muses were painted successively by Maccagnino and other artists, in- cluding Michele Pannonio and Cosmè Tura. Although one of Gazes’ most devoted students still remembered these verses over two decades later, around the time of Gazes’ death, they never seem to have been inscribed on the paintings that inspired them. When Ciriaco of Ancona visited Ferrara in 1449, the epigram he saw on the base of Clio was a Latin hexameter from Guari- no’s programmatic letter (1447). The painting of Thalia, com- pleted in 1456/7, after Leonello’s death, today also displays a Latin hexameter by Guarino, inscribed on a wooden tablet painted at the bottom, accompanied by a Greek hexameter translation. There is no trace of the cycle of epigrams written by Gazes and edited below. Urania, in the Pinacoteca Nazionale in Ferrara, inv. no. PNFe 399 (https:// gallerie-estensi.beniculturali.it/opere/collezioni/#/dettaglio/821427_La% 20Musa%20Urania). The lost panels were probably cut out when the paint- ings were removed from their original setting: see J. Anderson, “Il risveglio dell’interesse per le Muse nella Ferrara del Quattrocento,” in Le Muse e il principe 165–185, at 174; A. Bacchi, “La Musa Polinnia,” in Le Muse 408–416, at 410; D. Benati, “La musa Erato(?)” and “La musa Urania,” in Le Muse 383–395, at 383, 389–390; A. Mottola Molfino, “Le Muse dello studiolo: la diaspora, il collezionismo, il mercato, i restauri,” in Le Muse 223–233, at 223– 225; M. Natale, “La Musa Tersicore,” in Le Muse 395–404, at 398; Manni, Belfiore 85. The digital reconstruction (https://framelab.unibo.it/vt/ belfioretour/) implies the presence of similar tablets for each Muse. However, no traces of letters survive, so that the plates were either blank or the possible inscriptions and coat of arms were removed as a form of censorship or damnatio memoriae. Another Muse, now in the National Gallery in London, inv. no. NG3070 (https://www.nationalgallery.org.uk/paintings/cosimo-tura-a- muse-calliope) reveals what seems to be a portion of a painted piece of paper similar to the one that contains the name of the painter of Thalia. JULIÁN BÉRTOLA AND ANTHONY ELLIS 491 ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 It is uncertain whether Gazes wrote the whole cycle before he left Ferrara in 1449, that is, before all the paintings were finished and Leonello had died. But it seems that, in 1475, Ludovico Carbone could quote only the two epigrams corresponding to the first two paintings completed by 1449. Carbone indicates that the verses were composed at Leonello’s behest, but the nature of the request remains obscure. Did Leonello formally commission written verses for inscription on his paintings? Or did he simply desire an impromptu performance to celebrate a glorious occasion in a suitably erudite manner? Did Leonello request a cycle of poems for all nine muses, or only for the two paintings which had just been completed? While it seems natural to assume that Gazes completed the full cycle in Ferrara with Leonello’s sponsorship, it remains possible that Gazes wrote poems only for Clio and Melpomene during Leonello’s lifetime, and that he finished the rest at a later date, whether as a literary exercise, for entertainment, or as a set of poems which he could offer to friends or to another patron. Towards the end of Leonello’s life, Ciriaco of Ancona saw two paintings from the studiolo, but only one was captioned with Guarino’s verse. Ciriaco makes no mention of any Greek epi- gram—nor, for that matter, of any Greek text. But, as we learn from Carbone, by that time Gazes had already composed at least two poems for the same two Muses that Ciriaco saw. We also know for a fact that two wooden plaques were eventually painted below some of the Muses. This rather foggy scenario suggests that Leonello might have been considering various different options for furnishing his Muses with inscriptions. One might have been to complement Guarino’s Latin verses with some Greek poems, with each cycle displayed in its own plaque. At least one of Gazes’ students and supporters, Carbone, implored Leonello to have Gazes’ epigrams inscribed, presumably to the detriment of other candidates and no doubt echoing Gazes’ own aspirations. The situation suggests an environment of pro- fessional competition and intellectual tensions, if not outright 492 POEMS ON THE MUSES BY THEODOROS GAZES ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 rivalry.25 Ultimately, Gazes did not prevail. As we saw, after Leonello’s death, Guarino’s monostich was chosen for the Buda- pest Thalia, supplemented by a hexametric Greek translation.26 Nevertheless, Gazes’ cycle of epigrams was preserved in full, pre- sumably kept by Gazes in manuscript format, and the poems became part of the oral legend which students like Carbone remembered decades later. Within a few years of his death, the poems were in the hands of another member of the Greek diaspora in Rome, Demetrios Raoul Kabakes, who decided to 25 Carbone’s accounts reinforce this impression, as he remains oddly silent about Guarino’s verses. His request might be interpreted as asking Leonello to add Gazes’ poems to Guarino’s cycle, or to inscribe only Gazes’ verses— or even to inscribe Gazes’ verses accompanied by Carbone’s own Latin trans- lations. In this regard, it is noteworthy that Gazes’ epigrams do not differ drastically from the instructions of Guarino’s foundational letter of 1447. In general, Gazes follows an etymological interpretation of the names of the Muses, and thus they become allegories for their primary domains along the same lines as Guarino’s prescriptions: Clio for fame and glory, Euterpe for joy, Thalia for blossoming, Melpomene for singing, Terpsichore for dancing, Erato for love, Urania for (astronomical) knowledge, Calliope for the voice. Even the unexpected reference to sweat in Gazes’ poem on Polymnia can be better understood when read alongside Guarino’s poem which associates her with agriculture. Gazes also emphasizes the spiritual aspects of the Muses, albeit in a more balanced way than Guarino, through frequent parallelisms of body and soul, mortal and immortal, and so on. By contrast to Guarino, however, Gazes’ Muses do not seem concerned with aetiological narratives of inventions and discoveries. In Gazes’ elegiac couplets, the Muses speak entirely in the first person, whereas they do so in only some of Guarino’s hexametric monostichs. 26 It remains unclear why this happened. Perhaps Leonello wished to avoid offending Guarino and/or the scholar who had been asked to put the Latin hexameters into Greek—conceivably Guarino himself. Or perhaps Leonello was persuaded by Carbone’s request but, after his death and Gazes’ de- parture from Ferrara, the new plan was never executed, so that Leonello’s successors simply inserted Greek translations. It is worth mentioning another possible reconstruction: that Leonello formally commissioned Gazes’ poems for inscription and designed the plaque to house them, but for some reason abandoned or deferred that plan during his life. JULIÁN BÉRTOLA AND ANTHONY ELLIS 493 ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 copy them individually onto nine folia of his own codex of Herodotus.27 2. Herodotus and the Greek diaspora in Renaissance Rome We would not have Gazes’ poems today were it not for the fact that, since antiquity, each of the nine books of Herodotus’ Histories traditionally bears the name of a Muse: Clio, Euterpe, Thalia, Melpomene, Terpsichore, Erato, Polymnia, Urania, and Calliope. Some years after Gazes’ death, his poems on the Muses were reused for a new purpose, one which their author had probably never envisaged. Instead of being inscribed on the nine paintings of the Muses in the palace of Belfiore, they were used to decorate the openings of each of Herodotus’ nine books in a late-15th-century manuscript, to which we now turn. The poems are serendipitously preserved in Vat.gr. 1359, a copy of Herodotus’ Histories produced in Rome in 1480 by Demetrios Raoul Kabakes, a Peloponnesian aristocrat who had spent his life in service to the Byzantine Imperial family, and had been living in Rome for a little over a decade. The exemplar from which he made his copy of Herodotus was Florence, Biblioteca Medicea Laurenziana, Plut. 70.6, a manuscript which had been edited and in places rewritten by the controversial philosopher Gemistos Pletho. Kabakes’ copy of the Histories con- tains many marginalia and paratexts. Some of these were copied 27 As if anticipating the fate of the paintings of Belfiore and the rediscovery of Gazes’ epigrams, Carbone wrote the following before his quotation of the verses in Latin: “Unde pictorum vicem doleo, quod eorum tam egregia opera cuilibet minimo, vel aquae vel ignis, vel ipsius vetustatis quae omnia consumit periculo subiecta sint, at poemata bona et orationes cum dignitate conscripta immortalitatem et servant et praestant, ex unoque exemplari millia multa exemplorum quam primum fieri possunt atque in diversa loca transmitti” (“I pity the fate of painters, because their very excellent works are subject to any minor hazard of water or fire or aging itself, which destroys everything, whereas good poems and orations composed with merit preserve and confer immortality, and from one exemplar many thousands of exemplars can be made quickly and transmitted in different places”): Lazzari, Atti e Memorie 24 (1919) 35. 494 POEMS ON THE MUSES BY THEODOROS GAZES ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 from his exemplar,28 but most were added by Kabakes himself, apparently over the course of the 1480s.29 Among these are nine 28 For the textual history of Pletho’s manuscript, and the relationship between its many copies, including Vat.gr. 1359, see J. Bértola, “A First Critical Edition of the Cycle of Epigrams on Herodotus in the Margins of Manuscript Florence, Biblioteca Medicea Laurenziana, Plut. 70.6 and Some of Its Apographa,” JÖB 72 (2022) 63–96, and A. Ellis, “A Neo-Pagan Editor in Late Byzantine Sparta, Or: How Gemistos Pletho Rewrote his Herodotus,” DOP 78 (2024) 315–354. 29 On Vat.gr. 1359 see H. Stein, Herodoti Historiae I (Berlin 1869) XVII; E. Legrand, Bibliographie hellénique ou description raisonnée des ouvrages publiés par des Grecs aux XVe et XVIe siècles III (Paris 1903) 262; S. Lampros, “Λακεδαιµόνιοι βιβλιογράφοι καὶ κτήτορες κωδίκων κατὰ τοὺς µέσους αἰῶνας καὶ ἐπὶ Τουρ- κοκρατίας,” Νέος Ἑλληνομνήμων 4 (1907) 152–187, 303–357, at 332–333; A. Colonna, “De Herodoti memoria,” Bollettino del Comitato per la preparazione dell’ edizione nazionale dei classici greci e latini 1 (1945) 41–83, at 54; B. Hemmerdinger, Les manuscrits d’Hérodote et la critique verbale (Genoa 1981) 31–32; Bértola, JÖB 72 (2022) 74–75. The manuscript is today divided in two volumes (ff. 1–230 and 231–491, mm. 231 x 162), available online (https://digi.vatlib.it/view/ MSS_Vat.gr.1359.pt.1 and .pt.2). It also contains Lucian Herodotus or Aetion (ff. 1r–3r), and a letter from Kabakes to his son Manilius (ff. 487r–489r), on which see H. Lamers, “Manilius Cabacius Rhallus of Sparta (c. 1447–c. 1523): A Study of his Life and Work,” HumLov 62 (2013) 127–200, at 133– 134, and F. Bacchelli, “La Considération céleste et les Enseignements de Démétrius Rhaoul Kavàkis (avec deux lettres inédites de Gemistus Plethon),” Noctua 3 (2016) 164–238, at 169 n.14. For the annotations left by Kabakes in the mar- gins of Vat.gr. 1359 see the edition in preparation by A. Ellis, “A Renaissance Spartan Reads Herodotus: Demetrios Rhalles Kabakes’ Marginal Notes on Ancient Greek History in Vat. gr. 1359.” On Kabakes’ textual supplements to lacunae (copied into the margins of both Laur.Plut. 70.6 and Vat.gr. 1359) see D. Bianconi, “L’Erodoto di Nicola Tricline, Giorgio Gemisto Pletone e Demetrio Raoul Cabace. Il Laur. Plut. 70.6 da Tessalonica a Roma, pas- sando per Mistrà,” BollClass 43 (2022) 61–110. A monostich was copied at the end of Herodotus’ Histories in Vat.gr. 1359, f. 486v (ἰδὼν τὸ τέρµα τὴν χάριν θεῷ δίδου). This is a formulaic book epigram found in many other manu- scripts (see e.g. DBBE Type 2244), and used by Kabakes elsewhere (e.g. Vat.gr. 2237 and 2238). On Kabakes as a copyist see M. Vogel and V. Gardthausen, Die griechischen Schreiber des Mittelalters und der Renaissance (Leipzig 1909) 102; P. Canart, “Scribes grecs de la Renaissance. Additions et corrections aux réper- toires de Vogel-Gardthausen et de Patrinélis,” Scriptorium 17 (1963) 56–82, at JULIÁN BÉRTOLA AND ANTHONY ELLIS 495 ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 elegiac distichs, one dedicated to each of the Muses at the start of each of Herodotus’ nine books, and copied into the lower margin in red ink—the same ink used for the names of the Muses in the upper margin.30 Since Kabakes never mastered traditional Greek orthography and had no grasp of Greek metrical quantities, there is no question of his having written these verses.31 Their close corre- spondence to Ludovico Carbone’s Latin translations is, in itself, sufficient to identify them as the work of Theodoros Gazes. But Kabakes knew full well who their author was and noted his name in the margin next to the first epigram: Θεοδώρου τοῦ Γάζεω (“by 62, 77; D. Harlfinger, Specimina griechischer Kopisten der Renaissance I (Berlin 1974) 33; E. Gamillscheg et al., Repertorium der griechischen Kopisten (Vienna 1981, 1997) I no. 95, III no. 162; Bianconi 90–91. 30 At ff. 4r (Κλειώ), 56v (Εὐτέρπη), 118v (Θάλεια), 175r (Μελποµένη), 235r (Τερψιχόρη), 276v (Ἐρατώ), 320r (Πολύµνια), 395r (Οὐρανία), 441v (Καλλιόπη). 31 On Kabakes’ “orthographe fantaisiste” see D. A. Zakythinos, Le despotat grec de Morée. Edition revue et augmenté par Chryssa Maltézou II (London 1975) 375; on his “atrocious” spelling and “lowly” style, I. Ševčenko, “The Decline of Byzantium Seen through the Eyes of its Intellectuals,” DOP 15 (1961) 167– 186, at 174–175. For Kabakes’ own oblique references to the gaps in his education, see the remark in his γνῶµαι in Vat.gr. 2185, f. 33r–v (published by F. Bacchelli, “Di Demetrio Raoul Kavàkis e di alcuni suoi scritti,” Unomolti 1 [2007] 129–187, at 177–187, and most recently in Bacchelli, Noctua 3 [2016] 222–230): ἐπιδεὶ δὲ καὶ αὐτὸς ἦλθον παρὰ τῆς φύσεος πρὸς θεωρίαν κόσµου καὶ πρὸς τὸν βίον τὸν ἀνθρώπινον, ἄνθρωπος ὂν καὶ ἱκανὸς κατὰ χρόνον, καὶ καλὰ καὶ ἡ τίχη ἐπιστήµονα οὐδέν µε ἔπικεν ἢ ἀπὸ ἡµετέρας ῥαθειµίας ἢ καὶ ἐξ ἄλου τιχεροῦ τινός, ὅµος ὁµοίλισα µετὰ µεγάλλων καὶ σοφῶν ἀνδρῶν καὶ ἑλλήνων καὶ λατίνων καὶ πολλῶν βίβλων διήλθον πρὸς ἐξέτασιν καὶ ἐν τῖ Ἑλάδι καὶ ἐν Ἠταλίᾳ, ἐπειδὴ σχεδὸν πάσαι συνήχθησαν ἐνταῦτα µετὰ τὴν ἅλλοσιν τοῦ γένους (“When I departed from nature and came to the contemplation of the cosmos and to the condition of human life, being a man and of sufficient years—and fortune by no means granted me a good education, whether due to our own laziness or to some other circumstance—nevertheless I associated with great and wise men, both Greeks and Latins, and read many books both in Greece and in Italy, since almost all these had been collected here after the capture of my people”). 496 POEMS ON THE MUSES BY THEODOROS GAZES ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 Theodoros Gazes”).32 Fulvio Orsini (1529–1600), a later owner of the manuscript, seems to have taken this to mean that the whole codex had once belonged to Gazes.33 But this is certainly incorrect: the subscription at the end of the text makes it clear that Vat.gr. 1359 was copied by Demetrios Raoul Kabakes in Rome in the year that the Turks seized Otranto (1480), some five years after Gazes’ death.34 Kabakes was certainly well-placed to encounter Theodoros Gazes’ unpublished poetry. The two men probably never met during their youth in Greece, since Kabakes was doubtless very young when Gazes emigrated to Italy at some point in the 1430s.35 But in the decade or so between Kabakes’ arrival in Rome and Gazes’ death ca. 1475/6, these two distinguished émigrés moved in similar circles and probably knew one another well.36 32 Vat.gr. 1359, f. 4r. It is interesting that Kabakes uses the Ionic form of the genitive for Gazes’ name, since Ionic is the dialect used by Herodotus. 33 For Orsini’s description see P. de Nolhac, La bibliothèque de Fulvio Orsini (Paris 1887) 336: “Herodoto, che fù di Theodoro Gaza, libro integro, ligato in corame rosso, scritto in papiro in-4 foglio.” De Nolhac (146) understood this description to mean that this manuscript “provient d’une bibliothèque contemporaine de ce savant [sc. Gazes].” Vat.gr. 1359 also contains notes of possession by Orsini: βίβλος φουλβίου οὐρσίνου ῥωµαίου ἦν, εὗτε τάδ’ ἐγρά- φετο (f. 3v, “The book belonged to Fulvio Orsini, the Roman, when these words were written”) and τὸ παρὸν βιβλίον κτῆµά ἐστι, φουλβίου οὐρσίνου ῥωµαίου (f. 491r, “The present book is the possession of Fulvio Orsini the Roman”). 34 F. 486v: δηµητρίου ῥαοὺλ καβάκη· σπαρτιάτου καὶ βυζαντίου· ἐγράφη ἐν ῥώµη· ἐν ὧ χρόνω [ἔτει suprascripsit], ὄτροντω τούρκοι κατέλαβων (“By De- metrios Raoul Kabakes, Spartan and Byzantine. It was copied in Rome in the time [year] when the Turks seized Otranto”). 35 For Kabakes’ birth date see Bacchelli, Noctua 3 (2016) 169 n.13, and Ellis, in preparation: he was certainly born between 1398/9 and 1415/6, probably much closer to the latter date. 36 Due to his close links with the Imperial family, Kabakes arrived in Italy long after the fall of Constantinople in 1453 and the fall of the Despotate of Mistra in 1460. He seems to have remained in the northeast Aegean for a JULIÁN BÉRTOLA AND ANTHONY ELLIS 497 ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 One common link was Cardinal Bessarion. After his arrival in Rome, probably in 1466,37 Kabakes was in close contact with Bessarion and the Greek diaspora that congregated around him after the progressive fall of Greece to the Ottomans.38 In this phase of his life, Kabakes devoted much time to the copying of ancient Greek texts, and in several cases he seems to have copied manuscripts from Bessarion’s growing library.39 In the margins number of years after 1460 acting as governor of Lemnos and Imbros on behalf of Demetrios Palaiologos, who stayed in Greece as an Ottoman vassal, and received these two islands as an appanage from Mehmed II. On this period of Kabakes’ life see Bacchelli, Noctua 3 (2016) 179–184, and esp. the notes in Vat.gr. 173 (Strabo), as transcribed and interpreted by A. Diller, The Textual Tradition of Strabo’s Geography (Amsterdam 1975) 144: ταύτας [sc. τὰς νήσους] ἥρξα καὶ αὐτὸς, ἔχοντες αὐτὰς ἕλληνες, οὐ πάσας, ἀλλὰ λήµνον καὶ ἴµβρον· οὖπω γὰρ ἦν τελέως ἐφθαρµένον τὸ βασίληον τῆς ἑλλάδος (“I myself ruled these islands, when the Hellenes had them—not all of them, but Lem- nos and Imbros. For the kingdom of the Hellenes had not yet been entirely destroyed”). 37 Three subscriptions indicate the period in which Kabakes’ Italian life began: one was written “in Rome” in 1467 at the end of a copy of Aristotle’s Poetics (Vat.gr. 2238, f. 155r) and two were written in the “third year of Pope Innocent” (1487) in which Kabakes states that he has been in Rome for 21 years (Vat.gr. 173, f. 346v, and our Vat.gr. 1359, f. 491v), i.e. since 1466; see Bértola, JÖB 72 (2022) 74–75 n.62; Bianconi, BollClass 43 (2022) 63–64 n.10; Ellis, in preparation. 38 L. Mohler, Kardinal Bessarion als Theologe, Humanist, und Staatsmann I (Paderborn 1923) 249, 325–334; J. Monfasani, “Two Fifteenth-Century ‘Platonic Academies’: Bessarion’s and Ficino’s,” in Renaissance Humanism, from the Middle Ages to Modern Times (Aldershot 2015 [2011]) XIII. 39 For the link between Marc.gr. Z. 517 (coll. 886) (fragments of ancient authors copied by Pletho in a manuscript bequeathed by Bessarion to Venice) and Vat.gr. 2236 (an identical series of fragments copied out in Kabakes’ hand), see the detailed report in S. Lilla, Codices Vaticani Graeci. Codices 2162– 2254 (Vatican City 1995) 348–359. For Vat.gr. 2236 as an apograph of Marc. gr. Z. 517 and antigraph of Laur.Plut. 56.18, see A. Diller, “Pletho and Plutarch,” Scriptorium 8 (1954) 123–127, at 124 n.5, and “The Autographs of Georgius Gemistus Plethon,” Scriptorium 10 (1956) 27–41, at 31 n.14, 40. G. De Gregorio, “Attività scrittoria a Mistrà nell’ultima età paleologa: il caso 498 POEMS ON THE MUSES BY THEODOROS GAZES ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 of one of these manuscripts Kabakes records a dinner-time con- versation with Bessarion about the memory of the philosopher with whom they had both studied—Gemistos Pletho. This en- counter took place in Rome at some point between 1466 and the Cardinal’s death in 1472, and bespeaks close familiarity.40 Kabakes clearly commanded a certain prestige among his contemporary exiles: in 1472 he served as an ambassador to Moscow, representing the exiled Imperial family at a marriage which Bessarion had helped to broker between Ivan III of Russia del cod. Mut. gr. 144,” Scrittura e Civiltà 18 (1994) 243–280, at 269 n.71, has suggested that Vat.gr. 2236 was copied after Kabakes left Mistra (between 1447/8 and the end of the 1440s) but before he arrived in Rome in 1466, since the memories of Pletho found in the marginalia seem relatively fresh and the watermark suggests an earlier date (late 1450s or early 1460s). This date cannot be ruled out, but the arguments are not persuasive: some of its marginalia (see next note) were certainly written after 1466 since they record a discussion between Bessarion and Kabakes which took place “here” (ἐνταῦτα), i.e. in Rome, the only place where Kabakes and Bessarion could have met after the late 1430s; moreover, the spread of watermarks in Vat.gr. 2236 (not dated by subscription) is similar to those in Vat.gr. 2238, of which one section is dated to 1467 (Lilla, Codices 366, compares the watermarks to those found on other manuscripts written between 1459 in Naples and 1503 in Vienna). 40 Vat.gr. 2236, f. 141v in the lower margin: Ὁµηλοῦντος ἐµοῦ ἐνταῦτα περὶ τὴν σκολὴν τῆς τραπέζης µετὰ τοῦ ἐνδοξοτάτου γαρδυναλίου ἐκίνου κυρ. Βισαρίονος· ἐρέθει λόγος περὶ τοῦ Πλήθωνος· καὶ ἡρότισά τον ἐγῶ· “ἐµένη ἡ πρόληψις ῆν ὅριζες πολλάκις περὶ τοῦ Γεµηστοῦ, ἢ χαριζόµενος, τὰ ὅριζες.” ἀπεκρίθη ὅτι “οὐδὲν ἔλεγον χαριζόµενος, ἀλλὰ θέλο σε ἡπῆν µετὰ ἀλιθείας καὶ νῦν ὅτι ἀπὸ τοῦ Πλοτίνου τὸν κερὸν, ὃς ἦν πρὸ χιλίων τετρακοσίων ἐτῶν, σοφό- τερον ἄνθρωπον οὐδένα ἐποίησεν ἠ Ἑλλᾶς τοῦ Πλήθωνος.” Δηµήτριος (“When I was conversing here [i.e. in Rome] at the leisure of the table with that august Cardinal Bessarion, there arose a conversation about Pletho and I asked him: ‘Does that opinion which you often proclaimed about Gemistos abide, or were you merely being obliging when you proclaimed it?’ And he answered ‘I said nothing to be obliging, but I wish to tell you in truth even now that since the time of Plotinus, which was 1400 years ago, Hellas created no wiser person than Pletho’. Demetrios”). Text from G. Mercati, Opere Minore IV (Vatican City 1937) 173–174 n.2, reprinted with slightly different punctua- tion by Bacchelli, Noctua 3 (2016) 187 n.40. JULIÁN BÉRTOLA AND ANTHONY ELLIS 499 ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 and Zoe Palaiologina, the daughter of the late Thomas Palai- ologos (1409–1465) and niece of Demetrios Palaiologos (1407– 1470).41 During the same period, Theodoros Gazes was also a prominent member of Bessarion’s circle. After his fame as a teacher of Greek grew, Gazes left Ferrara in 1449/50 and went to Rome, where he enjoyed the patronage of Pope Nicholas V and worked on various translation projects from 1451 to 1455, during which time he was also in close contact with Bessarion. In 1456 he went to the court of Alfonso the Magnificent in Naples (where he stayed until 1458 or 1459). In the controversy on Plato, Aristotle, and Pletho which flared up in the late 1450s, Gazes collaborated with Bessarion in his polemical exchanges with George of Trebizond. And in his later years, Gazes enjoyed Bessarion’s direct patronage: in 1463 the latter appointed him to the benefice of San Giovanni a Piro in Calabria, which sup- ported him until his death. But in 1467, at the behest of Bessarion and Pope Paul II, Gazes returned to live in Rome close to Bessarion’s own residence in Santi Apostoli where he was viewed as the chief of Bessarion’s academy. On his return, he would have found Kabakes a new resident in the city.42 Gazes and Kabakes certainly knew one another personally, and some evidence of direct contact survives. Kabakes devel- oped a highly idiosyncratic theology based around the worship of the sun, and, prompted by this interest, he compiled a large collection of passages from ancient literature which mention the sun.43 One extract is a fluent Koine Greek translation of Pliny’s 41 R. Croskey, “Byzantine Greeks in Late Fifteenth and Early Sixteenth Century Russia,” in L. Clucas (ed.), The Byzantine Legacy in Eastern Europe (New York 1988) 33–56, at 37–39, 46–47. Kabakes’ marginalia on his crossing of the Alps are printed by Diller, The Textual Tradition 145. 42 See Bianca, in Dizionario Biografico LII 737–746. Note also the above- mentioned epigram by Gazes, dedicated to a manuscript of Homer which belonged to Bessarion, which also mentions the Muses (DBBE Type 5288). 43 On Kabakes’ helio-theology, and its uneasy relationship to the neo- Platonism of Pletho and other contemporaries, see especially Bacchelli, Noctua 3 (2016) 194–216. 500 POEMS ON THE MUSES BY THEODOROS GAZES ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 Latin, and Kabakes notes that the translator was Gazes.44 Kabakes was frank about his ignorance of Latin,45 and it seems likely that he asked Gazes directly for a translation, since the latter had worked alongside Andrea Bussi on an edition of Pliny’s Historia Naturalis, printed in Rome in 1470.46 Another connection between the two men was Kabakes’ son, Emmanuel (Ἐµµανουὴλ Ῥάλης Καβάκης). Although it was as a Latin poet that Emmanuel would gain renown—under the Latinized name Manilius Cabacius Rhallus—his Greek literary education47 owed something to Gazes: Manilius studied Gazes’ Greek grammar in a manuscript which he and his father had copied out.48 After Gazes’ death, around 1476, Manilius (not yet 44 The note on Vat.gr. 2185, f. 75v, reads: ἐρµϋνεῦς Θεόδωρος ὁ Γάζεος. See further P. Beullens and A. Gotthelf, “Theodore Gaza’s Translation of Ari- stotle’s De Animalibus: Content, Influence, and Date,” GRBS 47 (2007) 469– 513, at 491 n.70, and Bacchelli, Noctua 3 (2016) 188 and n.43. 45 At the end of his letter to Manilius describing their family history (copied at some point after 1480), Kabakes asks Manilius to translate the letter into Latin since he was unable to do so (Vat.gr. 1359, f. 489r): πρέπον δὲ καὶ αὐτὸς ὦ υἱέ µου, ποιήσε λατινηκὸς ταῦτα διὰ τὴν ἀγάπην τοῦ γένους. καὶ νὰ βωηθείσις καὶ τὸ σῶ πατρὶ ἀνεπιστήµονα ὄντα (“It is fitting for you, my son, to render this into Latin both for the love of your family, and so that you can help your father, who is ignorant”). 46 See here J. Monfasani, “The First Call for Press Censorship: Niccolò Perotti, Giovanni Andrea Bussi, Antonio Moreto, and the Editing of Pliny’s Natural History,” RenQ 41 (1988) 1–31, at 8 and n.28. On Gazes’ work on Bussi’s edition of Pliny see Beullens and Gotthelf, GRBS 47 (2007) 491–493. 47 For an assessment of Manilius’ Greek education—which must have begun during the progressive fall of the Byzantine world and presumably continued in Rome while he was acquiring the languages of his new home— see Bacchelli, Noctua 3 (2016) 187, and M. Manoussacas, “Cabacio Rallo, Manilio,” in Dizionario biografico degli Italiani XV (Rome 1972) 669–671, who speaks of a “solid literary formation” (669) in Mistra. 48 Vat.gr. 1378 contains Theodoros Gazes’ grammar with annotations by Manilius, who wrote a mark of possession on f. 1r: τοῦτο τὸ βιβλίον ἐστὶν ἐµοῦ Μανουὴλ Καβάκη τοῦ Ῥάλη. Hic liber est mei Emanuelis greci et cet. Cited from Bacchelli, Unomolti 1 (2007) 151 n.56, who thinks that the book played an JULIÁN BÉRTOLA AND ANTHONY ELLIS 501 ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 thirty) was invited to contribute to an anthology of poems com- posed in Gazes’ honor by Greek and Italian humanists.49 Whether the invitation reflects respect for his intimacy with Gazes, his poetic skills, or his aristocratic familial status, it seems likely that Manilius, like Kabakes, would have known Gazes well. We do not know how Kabakes came by a copy of Gazes’ cycle of poems to the Muses in 1480, almost half a decade after Gazes’ death. For all his social prestige, Kabakes can scarcely have cut an impressive literary figure in Gazes’ eyes. But he may have been an admirer of Gazes’ poetry and might have requested the opportunity to make a copy of the poems. Manilius is another possible route: the budding Latin poet and the aged Greek poet were probably aware of one another’s compositions and had friends in common, including the Neapolitan humanist Gio- vanni Pontano and Pomponius Laetus.50 Alternatively, the poems might have come to light after Gazes’ death in his papers, where they had sat since their composition over three decades earlier. Kabakes, an avid collector and compiler of Greek literature—ancient and contemporary—might have taken the chance to copy them before they were passed on to Demetrios important role in Manilius’ Greek education. Gamillscheg et al., Repertorium III no. 162, attribute ff. Iv–10v, 21r, 22r–28v, 51r–148r, 149v–192v to De- metrios Kabakes and, in Repertorium III no. 412, attribute ff. Ir, 29r–50v, and 149r to Manilius Kabakes. The composition of Gazes’ grammar (first printed in Venice in 1495 by Aldus Manutius) seems to date to his period in Ferrara— see Bianca, in Dizionario Biografico LII 739; for a later date of composition, see now P. Botley, Learning Greek in Western Europe, 1396–1529: Grammars, Lexica, and Classroom Texts (Philadelphia 2010) 14–16. 49 Lamers, HumLov 62 (2013) 160, 196; and now U. Mondini, “Greek and Latin Epigrams on the Death of Theodore Gaza,” in K. Kubina (ed.), Poetry in Late Byzantium (Leiden 2024) 191–235. 50 On these connections see Bacchelli, Noctua 3 (2016) 177–178, 187, who suggests that Manilius studied Latin in the school of Pomponius Laetus in Rome (who himself had studied with Gazes in the 1450s) and that he studied Greek with Gazes after 1467. There is, however, no direct evidence for this reconstruction. 502 POEMS ON THE MUSES BY THEODOROS GAZES ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 Chalkokondyles, to whom Gazes left all his books with only a few exceptions. The fact that the first letter of many poems seems to be missing suggests that the copy Kabakes encountered was a calligraphic version awaiting rubrication or illumination—and that the accuracy of Kabakes’ copy was not checked by the poems’ author. 3. This edition The only known Greek witness for our poems is Vat.gr. 1359. The text is generally copied with reasonable accuracy; a few imprecisions can be emended without much trouble (e.g. the breathings of οὕνοµά 1.1 and οὕτις 8.2, οὐρανία and ἀθάτοι in 8.1; the case of τέρψαµόνη in 5.2 is more ambiguous: as it seems, it was corrected by Kabakes into τέρψαµένη, but the double ac- cent and the general sloppiness of the word makes us wonder whether Gazes could have written τερψάµενοι or even τερψα- µένην, instead of τερψαµένη). Another obvious correction is the degemination in ἄλους (2.1). Degemination is a common phe- nomenon in medieval Greek, but, Kabakes’ spelling mistake aside, the alpha needs to be long and the double lambda makes position. The metre of Gazes’ elegiac couplets is generally correct, with mostly penthemimeres caesuras in the hexameters—except for 3.1 (hephthemimeres) and 6.1 (trihemimeres and bucolic cae- sura)—and the regular caesura in the middle of pentameters. No vowel or diphthong is given the wrong length, but in 5.2 we have corrected κᾱρδίην to κρᾰδίην, which fits the meter and avoids an odd correption.51 On the other hand, as in Homer, correptions are avoided in the contiguous vowels in αἰδ ο̄̄ί η (5.1) and in Μουσά̄ων (6.1; cf. ἀδελφά̄ων 8.2 after this model). Gazes made generous use of elision (e.g. 1.1, 2.1, 3.2, 4.1) and of particles (e.g. 6.2, 8.2) to avoid hiatus. In 5.1, we have also added the enclitic 51 Kristoffel Demoen and Arnd Kerkhecker have independently suggested this conjecture. There is also correption in 4.2 (θέλξᾰ βροτοὺς), but this is a more common case of Attic correption with a voiced plosive. JULIÁN BÉRTOLA AND ANTHONY ELLIS 503 ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 accent in εἶπέ γε.52 It seems likely that the initial letters of every poem were left blank in the exemplar from which Kabakes made his copy, and that he only supplemented the evident cases, that is, when the poems start with the names of the Muses: τερψιχόρην 5.1, οὐρανία 8.1, καλλιόπη 9.1; probably µελποµένη 4.1 as well, but there the manuscript has a lacuna. In the other cases, he supplied the first words with breathings and accents to make some sense of them: ἔρποµ’ 2.1, θρεῖ 3.1, ἴµ’ 6.1, εἶνε 7.1.53 The case of 1.1 is less clear, as both οὔνοµά and τοὔνοµά are possible. In the edition we have sought to respect the punctuation of the manuscript, though we have introduced some simplifications. At the end of the first verse of each poem, punctuation is often present in the manuscript (raised dot in 2.1, 6.1, 8.1, 9.1; no punctuation in 3.1), but we have sometimes changed it (1.1, 4.1, low dot to a raised dot; 5.1, low dot to comma; 7.1, colon to raised dot). Within the verses, some commas are already present in the manuscript (3.2, 4.2, 6.2), some have been removed (be- 52 This was a suggestion by Arnd Kerkhecker. Although the Byzantine practice of accenting enclitics does not always follow modern conventions for ancient Greek accentuation, we have chosen to normalize the enclitic accent here. In Laur.Plut. 32.1, f. 605v, Gazes himself copied Iliad 23.461: κεῖσέ γε. We have also added the enclitic accent in 8.1 ἀθάνατοί (see Il. 23.277 in Laur.Plut. 32.1, f. 598v). We have silently normalized some accents (e.g. 2.1 φρέν’, where the manuscript seems to read φρὲν’, and 6.1 Ἐρατὼ, where the manuscript seems to read Ἐρατώ). However, we have refrained from changing the grave accent of oxytone words to an acute before punctuation (4.2 βροτοὺς) following the common practice in Byzantine manuscripts, also widely attested in manuscripts copied by Gazes. 53 Some of these corrections often force farfetched interpretations: ἔρποµ’ could be taken as if coming from ἕρποµαι (but note the wrong breathing), θρεῖ as if from θρέοµαι (very rare in active voice). Other interpretations are facili- tated by common pronunciation of vowels and diphthongs: ἴµ’ as if from εἴµ’ and εἶνε as if from εἶναι. They suggest that Kabakes was not, himself, fully able to understand Gazes’ poetic productions in archaic dialects. In any case, Kabakes always supplemented breathings or accents to render the words acceptable to the eye. 504 POEMS ON THE MUSES BY THEODOROS GAZES ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 fore εὔχοµαι 7.1, before and after γνῶσιν 8.2, before καὶ 9.2), one has been added (2.2, no punctuation), and several raised dots have been changed to commas (1.2, 3.1; a dash to a comma at 7.1). A full stop has been written at the end of every poem, where the manuscript has either a colon (1.2, 2.2, 3.2, 5.2, 6.2, 7.2, 8.2, 9.2) or a raised dot (4.2). Additionally, some poems have a colon before the start (5.1, 6.1, 9.1), which has been removed. The first letter of every poem and the initials of the names of Greek deities have been written in capitals. The only case of mute iota in our cycle is notably written on the line in the manuscript, as iota adscript (ἔχηις 3.2). It seems likely that this is connected with the epigraphic framing which the author originally—and perhaps over-optimistically—envisaged for the cycle.54 The folio containing poem 4 is damaged and the beginning of both verses has been obliterated. Whereas the first verse is easy to reconstruct (Μελποµένη 4.1), the second leaves more room for conjecture. Fortunately, it corresponds to one of the two poems 54 If that is the case, we could venture that Kabakes’ exemplar was written in Greek majuscules. Adscript iotas are not unexpected in Greek inscriptions in majuscules on Renaissance paintings: see e.g. Sandro Botticelli’s Mystic Nativity, discussed by Wilson, IMU 35 (1992) 232–241. Wilson (234) rightly points out that the peculiar way of writing the iota adscript in this painting (on the line, but in a smaller size than the rest of the letters) resembles the typeface used by the printer Lorenzo de Alopa in collaboration with Janus Lascaris (see e.g. the Greek Anthology printed in Florence, 1494). However, this feature is also found in Gazes’ autograph copy of the Batrachomyomachia and the Iliad preserved in Laur.Plut. 32.1, dated early in the 1440s, when Gazes was in Pavia. As noted by S. Rizzo, “Gli umanisti, i testi classici e le scritture maiuscole,” in C. Questa et al. (eds.), Il libro e il testo (Urbino 1984) 225–241, at 237, the beginning and the end of this manuscript show a more distinctive majuscule script, and thus the mute iota is written on the line in a smaller size: see e.g. Batrach. 32 ποιῇ in f. 2r (= Speranzi, Medioevo e Rinascimento 23 [2012] pl. II) and Il. 24.772–774 σῇ, τῷ, Τροίῃ εὐρείῃ in f. 653v. A. Pontani, “Le maiuscole greche antiquarie di Giano Lascaris. Per la storia dell’alfabeto greco in Italia nel ’400,” Scrittura e Civiltà 16 (1992) 77–227, at 118, lists iota adscript among the characteristics of the antiquarian majuscules in mid-15th century Italy, in whose development Ciriaco of Ancona played an important role. JULIÁN BÉRTOLA AND ANTHONY ELLIS 505 ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 recorded by Carbone in Latin. When translating more literally, in prose, Carbone renders the partially lost word as “cantu.” Elsewhere in Carbone’s prose translation, in the poem for Clio, ἀείδω (1.1) is the word rendered as “canto.”55 The conjecture ᾄσµασι for 4.2 thus seems quite likely.56 In the versified version of his translation, Carbone renders the missing word as “voce,”57 which might correspond to φθέγµασι. But since the prose transla- tion is more faithful to the original than the elegiac translation, the former seems preferable. Besides, “voice” seems rather to belong to the domain of Calliope (cf. φθέγµα, 9.2). In retroverting the Greek verses from Latin, Wilson conjectured ὀπί as the original behind “voce,” but the ending -µασι, still visible in the new manuscript evidence, rules this out. The critical apparatus is generally negative, meaning that we record the readings of the manuscript whenever we do not follow them, because we print our emendations or conjectures—more uncertain conjectures are indicated with “fortasse” in the ap- paratus. It is positive only when we are not entirely sure of the reading of the manuscript—such cases are indicated with “ut videtur.” For poems 1 and 4, the apparatus includes a small selection of relevant references to Carbone’s Latin translations (in prose and in verse) and Wilson’s retroversion of them into Greek, which was based solely on Carbone’s text. The references 55 Lazzari, Atti e Memorie 24 (1919) 35. 56 Before the authors became aware of Carbone’s translation, this con- jecture had been suggested by Antonia Apostolakou. As for the iota subscript in ᾆσµα, the most extended practice in Byzantine scribes was not to add it to the root of words: see e.g. Gazes’ autograph copy of Plato’s Republic 424C in Laur.Plut. 80.7, f. 43v, where ἄσµατα is copied without iota subscript (else- where marked with a small dot below the vowel in this MS.). However, we have opted to include the iota subscript, abiding by the established conven- tion, for lack of any further evidence in these poems. Originally, there could have also been an iota adscript (as in ἔχηις 3.2), or an iota subscript, even if Kabakes’ exemplar was written in majuscule (see the title in Laur.Plut. 32.1, f. 17r, ῥαψῳδίας). 57 Lazzari, Atti e Memorie 24 (1919) 37. 506 POEMS ON THE MUSES BY THEODOROS GAZES ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 to Carbone and Wilson are limited to specific loci desperati and should not be understood as conjectures—the apparatus is neither positive nor negative in these cases. 4. The poems Sigla V = Vat.gr. 1359 (anno 1480) (...) = lacuna codicis Carbone = translatio latina Ludovici Carbonis Carbone2 = translatio latina Ludovici Carbonis (versio poetica) Wilson = coniecit in versibus graecis Nigel Wilson 1. Κλειώ Οὔνοµά µοι Κλειώ κλέος ἀνδρῶν οὕνεκ’ ἀείδω· χρύσεον εἷµα φέρω, χρύσεον ἦτορ ἔχω. 1 Οὔνοµά µοι] οὕνοµά µοι V, οὔνοµ’ ἐµοὶ Wilson, fortasse τοὔνοµά µοι | κλέος V ut videtur Clio My name is Clio because I sing the glory of men. I wear a golden garment and have a golden heart. 2. Εὐτέρπη Τέρποµ’ ἀεὶ φρέν’ ἐγὼ τέρπω τ’ ἄλλους Εὐτέρπη· κόσµιος εὐφυίαν, κόσµιος εὐστοµίαν. 1 Τέρποµ’] ἔρποµ’ V | ἄλλους] ἄλους V Euterpe I, Euterpe, always delight in my heart and delight others, seemly in my shapeliness and seemly in sweet sound. 3. Θάλεια Ἄθρει µου δέµας ἀνθηρᾶς, ὅ τέθηλε, Θαλείας ἄνθεά τε ψυχῆς, νῦν βλέφαρ’ εἴπερ ἔχῃς. 1 Ἄθρει] θρεῖ V || 2 νῦν V ut videtur | ἔχῃς] ἔχηις V Thalia Behold my body, that of blossoming Thalia, which has come into bloom, and behold the flowers of my soul, if only you have eyes to see. 4. Μελποµένη Μελποµένη ξέν’ ἐγὼ κούρη Διὸς αἰὲν ἐόντος· ᾄσµασι θέλξα βροτοὺς, ὄµµασιν ἀθανάτους. JULIÁN BÉRTOLA AND ANTHONY ELLIS 507 ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 1 Μελποµένη] (…)ποµένη V, Μελποµένη Wilson || 2 ᾄσµασι] (…)µασι V, cantu Carbone, voce Carbone2, ὀπὶ Wilson Melpomene Stranger, I am Melpomene, daughter of everlasting Zeus. I enchanted mortals with my songs and immortals with my looks. 5. Τερψιχόρη Τερψιχόρην µε πατὴρ αἰδοίη τ’ εἶπέ γε µήτηρ, βήµασιν εὐρύθµοις τερψαµένη κραδίην. 1 εἶπέ] εἶπε V || 2 τερψαµένη] τέρψαµόνη ante correctionem, τέρψαµένη post correctionem V ut videtur, fortasse τερψάµενοι vel τερψαµένην | κραδίην scripsimus metri causa, καρδίην V Terpsichore Terpsichore my father called me, as did my venerable mother, delighted in her heart by my rhythmical steps. 6. Ἐρατώ Εἴµ’ Ἐρατὼ Μουσάων ἱρῶν εἶδος ἀρίστη· ἀθάνατος δὲ νέων, ὅστις ἐµοῦ γ’ ἐράοι. 1 Εἴµ’] ἴµ’ V Erato I am Erato, most beautiful in aspect among the holy Muses. Among the young, whoever loves me is immortal. 7. Πολύµνια Ξεῖνε, Πολύµνια µὲν θυγάτηρ Διὸς εὔχοµαι εἶναι· στέµµατα δ’ ἱδρῶσι χρύσεα βάψα τάδε. 1 Ξεῖνε] εἶνε V Polymnia Stranger, I proudly declare that I am Polymnia, the daughter of Zeus. I moistened these golden garlands with sweat. 8. Οὐρανία Οὐρανίαν µε καλοῦσι θνητοί τ’ ἀθάνατοί τε· οὔτις ἀδελφάων γνῶσιν ἔµοιγ’ ἐρίσει. 1 Οὐρανίαν] οὐρανία V | ἀθάνατοί] ἀθάτοι V || 2 οὔτις] οὕτις V Urania Mortals and immortals call me Urania. None of my sisters will contend with me in knowledge. 508 POEMS ON THE MUSES BY THEODOROS GAZES ————— Greek, Roman, and Byzantine Studies 64 (2024) 480–508 9. Καλλιόπη Καλλιόπη πέλοµαι Διὸς εὐειδέστατον ἔρνος· φθέγµα δέ µου στόµατος καὶ µέλιτος γλύκιον. Calliope I am Calliope, most beautiful scion of Zeus. The voice from my mouth is sweeter even than honey.58 March, 2024 Seeger Center for Hellenic Studies Istituto Svizzero di Roma Princeton University Via Ludovisi 48 Scheide Caldwell House 00187 Rome, Italy Princeton, NJ 08544 U.S.A. anthony.ellis@unibe.ch jb0278@princeton.edu 58 This article is the result of the joint effort of its authors. The poems were discovered independently in the context of Ellis’s research on the marginalia of Vat.gr. 1359 and Bértola’s dissertation on verse scholia and book epigrams on historians; in this context, Bértola added the poems to DBBE. In this article, Bértola wrote most of section 1 and Ellis most of section 2, while the edition itself has been produced by both authors. Bértola has mainly written his share as a postdoctoral fellow of the Research Foundation – Flanders (FWO) at Ghent University. We are grateful to Arnd Kerkhecker for his inspired suggestion that many poems appear to be missing an initial letter, which has proved extremely helpful in the reconstruction of many verses, and for his careful comments on the entire paper. Our thanks also go to Antonia Apostolakou, who has helped us with the conjecture in poem 4. Ugo Mondini generously shared with us his then unpublished chapter on the funerary epi- grams on Gazes’ death. Kristoffel Demoen and Filippomaria Pontani read earlier versions of this paper and provided very useful suggestions. Finally, this article greatly benefited from the comments by editors and the anon- ymous reviewers.