Hostname: page-component-74656f9884-8r8nv Total loading time: 0 Render date: 2026-11-06T14:17:39.719Z Has data issue: false hasContentIssue false
  • English
  • Français

THE DESCENDANTS OF PETRARCH'S PRO ARCHIA*

Published online by Cambridge University Press:  24 April 2013

Jeroen De Keyser
Jeroen De Keyser*
Affiliation:
University of Leuven
*
Jeroen.DeKeyser@arts.kuleuven.be
Rights & Permissions [Opens in a new window]

Extract

Editors of Cicero's Pro Archia have assumed that Petrarch's lost transcription of the equally lost Liège manuscript that he discovered in 1333 survives in an almost unaltered version in a single Florentine manuscript, while the remaining 265 Itali reflect another stage of the text, when conjectural corrections by its learned discoverer were introduced into the text. This article proposes a reassessment of that dichotomy, based on a first comprehensive study of the whole transmission.

Information

Type
Research Article
Information
The Classical Quarterly , Volume 63 , Issue 1 , May 2013 , pp. 292 - 328
Copyright
Copyright © The Classical Association 2013

Editors of Cicero's Pro Archia have assumed that Petrarch's lost transcription of the equally lost Liège manuscript that he discovered in 1333 survives in an almost unaltered version in a single Florentine manuscript, while the remaining 265 Itali reflect another stage of the text, when conjectural corrections by its learned discoverer were introduced into the text. This article proposes a reassessment of that dichotomy, based on a first comprehensive study of the whole transmission.

PETRARCH'S USE OF HIS DISCOVERY

All but four of the almost 270 extant manuscripts of Cicero's Pro Archia derive from the transcription that Petrarch made in 1333 in Liège, where he discovered an exemplar of the speech along with another, still unidentified Ciceronian oration.Footnote 1 That summer, the twenty-nine-year-old Petrarch sent two letters to Cardinal Giovanni Colonna (Fam. 1.4–5), describing his visits to Paris, Ghent, Liège and Aachen chiefly from a touristic angle. Only four decades later did he describe in detail his find, relating in another letter how he had heard about a rich library in Liège, and then together with a friend had copied two Ciceronian speeches after first having experienced some difficulty in finding decent ink.Footnote 2 Unfortunately, both the Liège exemplar and Petrarch's copy have been lost. Notwithstanding the enormous importance of these lost testimonia – both for reconstructing the archetype and for outlining the rapid proliferation of the speech from that moment, all over Italy and then throughout the rest of Europe – no real attempt has ever been made to classify in any detail the prolific progeny of Petrarch's find.

Proud as he may have been of his discovery, Petrarch – who was notoriously parsimonious when it came to sharing his finds – seems to have sat on it for quite some time. The first indications of the reception of the Pro Archia in Italy are indeed Petrarch's own writings. When, in 1341 on Rome's Capitoline Hill, he received the crown of ‘poeta laureatus’, he pronounced his Collatio laureationis, an oration that argues at its core for the sacred character of poetry. Both the concept and its phrasing can be traced back to the speech that he had discovered a few years before in Liège.Footnote 3 He even refers explicitly to the source of his inspiration, claiming Cicero's literary authority: ‘Do not believe me, but Cicero, who in his oration for Aulus Licinius Archias speaks about poets in the following terms …’Footnote 4

When it comes to the propagation of the actual text, our only certain knowledge is that almost two decades after his discovery, Petrarch sent a copy to his Florentine friend Lapo da Castiglionchio.Footnote 5 We have the covering letter Petrarch sent with his gift to Lapo, written in Parma, probably in January 1351, in which Petrarch asks to be rewarded in exchange with three Ciceronian speeches he had seen the year before at Lapo's house in Florence, when he was on his way to Rome on the occasion of the Jubilee,Footnote 6 as Lapo's library contained a codex with speeches absent from his own collection. The codex actually contained not three but four speeches: Pro lege Manilia, Pro Milone, Pro Plancio and Pro Sulla, as can be deduced from Petrarch's letter of thanks to Lapo, sent a few months later, and from a later letter to him.Footnote 7 Petrarch kept Lapo's loan for almost four years, claiming – when he finally sent it back – that he had not been able to find a competent copyist, and in the end had transcribed the text himself. In the meantime, however, he had written in 1352 to another of his Florentine friends, Francesco Nelli, that he could deduce from allusions in the letters that arrived from Florence that the Pro Archia was being circulated and attentively read there.Footnote 8

LAPO'S ALLEGED MANUSCRIPT

For quite some time, scholars have assumed that material proof of the exchange between Lapo and Petrarch survived: because of erroneously read marginal notes in manuscript Plut. 23 sin. 3 of the Biblioteca Medicea Laurenziana (a), Albert C. Clark assumed that this Ciceronian collection had been transcribed for, if not by, Lapo,Footnote 9 and as the codex contains among others all five speeches involved in the exchange with Petrarch, it was believed to be Lapo's own copy of the four orations that he sent to Petrarch and of the one he received in exchange.Footnote 10 Furthermore, since in a the Pro Archia is immediately followed by the spurious Pridie quam in exilium iret, it seemed obvious to Clark that this was the unnamed other speech Petrarch had come across in Liège.

Giuseppe Billanovich surmised that Petrarch left the second discovery unmentioned in his letters because he considered it an irrelevant forgery. However, it appears that Petrarch did not refrain from sending it paired with the Pro Archia to his Florentine friends. Furthermore, Billanovich argued that a cannot be considered Lapo's own copy, and considered it a direct descendant of it.Footnote 11 He maintained a's connection to Lapo and its unique position in the whole exchange with Petrarch, but in 1991 Silvia Rizzo disproved the presence of Lapo's hand in a, adhering, however, to the bifurcation between a and all other Itali.Footnote 12 The pertinent objection that ‘there is a puzzle in this [bifurcation] that Petrarchan scholars have not noticed: Petrarch had had Pro Archia since 1333, and yet the text that he gave Lapo nearly twenty years later had none of his emendations’Footnote 13 is answered by Rizzo with the suggestion that Petrarch either deliberately did not copy his corrections and annotations and include them with the text he sent to Lapo, or that he undertook his philological effort only upon receiving Lapo's texts and integrating them in his own collection.

Yet another contradiction has, in my opinion, never been addressed. Nobody seems to have taken exception to the fact that, while the sequence of Pro Archia and Pridie was supposed to be an indication of the identity of the second text discovered by Petrarch in Liège, strangely enough Lapo's quartet is displayed by a in a rather disorderly way: Pro Milone, Pro Plancio and Pro Sulla conclude the manuscript, but Pro lege Manilia is oddly enclosed between the three Caesarian speeches (Pro Marcello, Pro Ligario and Pro rege Deiotaro) and the Pro Archia, followed by Pridie, Red. sen., and only then do the three other members of the quartet conclude the manuscript.

THE TRANSMISSION OF CICERO'S PRO ARCHIA

Before entering at more length into an examination of the numerous offspring of Petrarch's copy, it may be useful to rehearse briefly the prevailing views among modern editors about the relationship between the major testimonia of the Pro Archia. The stemma printed by Helmut Kasten, the most recent critical editor of the text, displays a first branch consisting of Lat. fol. 252 of the Staatsbibliothek in Berlin (E, c. 1158) and its late twin Palat. lat. 1525 (V, copied in 1467 in Heidelberg), and a second branch represented by the oldest manuscript (G), probably written in the eleventh century for Olbert, Abbot of Gembloux, now cod. 5348–52 of the Koninklijke Bibliotheek in Brussels, which is generally also considered the codex optimus of the text, and its sibling, the lost Liège manuscript or Leodiensis. From the manuscript found by Petrarch in Liège descend the already mentioned a (Plut. 23 sin. 3) on the one hand, and ϕ, allegedly the source of all other descendants of Petrarch's copy, on the other – the consensus of a and ϕ usually being rendered in the apparatuses as ω.Footnote 14 As the evidence for uniting Petrarch's model with G against E and V is rather weak, however, one might also deduce from the variae lectiones as they are displayed in modern editions a tripartite stemma, and it has even been suggested that the Leodiensis was the archetype of G and E + V.Footnote 15

At any rate, the bifurcation between a and all other Italian witnesses of the newly established Ciceronian quintet, descending from one Petrarchan codex (ϕ), in which the copy of Lapo's loan merged with the Pro Archia brought back from Liège, as was suggested a century ago by Clark in his Oxford edition of the speeches, has not been challenged since. Nor has anyone really widened the range of testimonia on which Clark founded his judgment. Yet even without undertaking extensive collation, one can surmise that the improbable bifurcation between one single manuscript and ‘all others’ is all the more likely to arise when the sample representing ‘all others’ consists of barely half a dozen manuscripts out of a mare magnum of more than 260. Furthermore, one cannot but conclude from a quick glance at any page of a that it displays a degree of corruption to an extent that none of the editors' apparatuses would lead one to suspect.

A fresh look at the evidence therefore seems appropriate in order to get a better grasp of the transmission. The present assessment is based on an examination of 249 of the 268 extant manuscripts I have been able to find, and aims at a more accurate reconstruction of Petrarch's copy (which I will call π) and hence a better assessment of its relation to the other three stemmatically relevant witnesses, G, E and V.

As I have already hinted, another of Clark's claims – that of the existence of a familia Gallica, occupying a middle position between the Itali and the older Northern witnesses G and E – was rebutted by Billanovich as long ago as 1946.Footnote 16 However, as Silvia Rizzo pointed out in 1975,Footnote 17 Billanovich's findings seem to have escaped the notice of editors of Cicero's speeches for at least three decades. Billanovich indeed found the so-called familia Gallica of numerous Ciceronian speeches to be nothing else than a branch of the tradition in which numerous sound readings are in fact emendations by Petrarch that had been incorporated in the text. Billanovich and Rizzo identified two valuable old Vatican codices in which Petrarch's conjectures have been passed down as marginal notes. Billanovich first discussed Palat. lat. 1820 (P), while Rizzo subsequently pointed to Vat. lat. 9305 (Q, c. 1375–1400), which contains an even more extensive list of marginalia.Footnote 18 In the former, copied in 1394 at Padua, the Pro Archia is preceded by the four speeches that Lapo provided Petrarch with, and it is followed by the subscription Marcii Tullii Ciceronis quinque orationes preclarissime expliciunt, which is deemed to be a clear indication of the five forming a well defined unit in the exemplar. Almost the same subscription is found in an early fifteenth-century manuscript, W241 of the Biblioteca Comunale in Trent (T), which has Expliciunt quinque M. Tulli Ciceronis orationes preclarissime. Ezio Ornato, who first called scholars' attention to this striking correspondence, observed, however, that the copyist of the Trent manuscript had left out Petrarch's marginalia.Footnote 19 As a matter of fact, he did not: he – or one of his predecessors –incorporated them without any distinction in the main text, producing thereby some rather nonsensical sentences. Philological analysis proves T – to my knowledge the sole manuscript containing only the canonical ‘Lapian’ quintet and nothing else – to be close to P and even closer to Q, where the Pro Archia however does not conclude the manuscript, since Q additionally contains Pro Marcello, Pro Ligario and Pro rege Deiotaro (and then Pro Cluentio, Pro Quinctio, Pro Flacco and the first part of Red. pop.).Footnote 20 The same sequence – first Lapo's quartet, then Pro Archia followed by the three Caesarianae – can be found in more than fifty other codices. This recurrent order in itself does not help much to prove the precise stemmatic relations among all these manuscripts, or to identify the more important members of the family, but neither should its importance be underestimated. In Antiquity most of Cicero's speeches were first published separately, and then circulated in collections of modest scope. During the Middle Ages several collections of speeches, mostly in random order, were assembled, but only in the fifteenth century, after the discovery of several speeches that hardly circulated during the Middle Ages, was there a general effort at producing compilations of as many of Cicero's orations as possible.Footnote 21 Given the fact that we know for sure that all but three copies of the Pro Archia originated from Petrarch's portfolio, and that they were all produced in a relatively limited time frame, between the discovery of the speech in 1333 and the advent of the printing revolution in the 1470s, a first taxonomy based on the recurrence of identical sequences in the packages that contain our text may help in producing a more detailed picture of the transmission.Footnote 22

Clark based what later came to be ϕ – the branch to which P, Q and T according to current knowledge may be deemed to belong – on P and half a dozen other manuscripts. All subsequent editors followed in his footsteps. Gaffiot's Budé text can hardly be considered a real critical edition: in his apparatus he only cites G and E, reducing all other witnesses to an all-embracing rec. scr. (‘recentior scriptura, id est emendatio aliqua recentioribus codicibus prolata quos significare nominatim supervacaneum videtur’), and in his preface he simply lists five of Clark's manuscripts, dismissed as having no other interest than ‘de présenter çà et là les corrections de quelque érudit inconnu’, without substantiating his claim.Footnote 23 Kasten's main contribution is his detailed report of the particular errors of E, as he seems to rely on Clark's collations for all other witnesses (and at any rate sticks to the very same manuscripts).

β: A DISTINCT NEW FAMILY

The question of how representative Clark's handpicked testimonia to ϕ really are seems never to have been addressed by any of his successors – not even after the refutation of his familia Gallica thesis – and the unique position of a as the only manuscript independent of ϕ within ω has never met with any opposition in editions of other Ciceronian speeches either. Although Q and T were not used by any of the modern editors (P is quoted in a few loci as p by Clark and Kasten), their readings do often correspond to the ones reported by Kasten as ϕ. Even a slightly broader collation of some of the older testimonia, however, points to the existence of a very large group of manuscripts (from now on β) that present neither a's peculiar readings nor all the omissions and obvious errors of ϕ. Moreover, some fifty of them have in common three corruptions that set them apart from the rest of the tradition:Footnote 24

  1. 14.19 scriptores ] scriptiones β

  2. 23.11 minus ] om. β

  3. 32.29 probata esse ] esse probata β

Within β, a distinction can be made between manuscript D 69 inf. of the Biblioteca Ambrosiana in Milan (B), a relatively old codex (c. 1391–1410, which makes it one of only half a dozen π codices that may predate the year 1400), and all other members of β, which I will call β1. The uncontaminated members of this predominantly Florentine class display a much longer series of common errors:

  1. 8.11 et fide ] ac fide β1

  2. 11.11 superioribus cum eodem quaestore fuisse] om. β1

  3. 12.22 defessae ] om. β1

  4. 13.7 quoque ] om. β1

  5. 15.23 effers laudibus ] laudibus effers β1

  6. 16.6 ceterae ] certe β1

  7. 17.11 gustare possemus ] om. β1

  8. 18.23 vero ] nec β1

  9. 19.5 oppido dedicaverunt ] urbe dedicarunt β1

  10. 19.7 post mortem etiam ] etiam post mortem β1

  11. 20.14 Athenis virum ] virum Athenis β1

  12. 21.22 ipsa ] om. β1

  13. 29.28 terminaret ] determinaret β1

  14. 30.6 statuas ] cum statuas β1

  15. 31.23 domesticis ] om. β1

  16. 31.26 violatus ] om. β1

These errors are absent from B and from the rest of tradition, with the exception of the interpolation cum at 30.6, which proliferated by contamination.Footnote 25

One of the oldest and most interesting members of this β1 family is the Biblioteca Laurenziana's Plut. 76.32 (F). Apart from the obvious β1 errors, it also presents some peculiar readings – emendations, as it seems – that subsequently surfaced again only in a few relatively late fifteenth-century Ciceronian collections that are the product of wide compilation and philological erudition, and that present a synthesis of the better readings present in the wide open field of Itali.Footnote 26 At 10.7, for instance, F reads the ne utitur that Clark drew from his manuscript c (Bodleian, Canon. Class. 226, a compilation from the middle of the fifteenth century).Footnote 27 Furthermore, at 15.24 it has the editors' est certum quod against the est certum quid of almost all other Itali; and at 19.4 it reads Smirnei (for Smyrnaei), while almost all other manuscripts – G, E and V included – have Smirnii or an even more corrupt alternative. These emendations are absent from B and the rest of β1 yet there are no traces of any philological intervention on F itself. Nevertheless, F must be rather close to the actual manuscript β1, as it is the only member of the subfamily to display a gap instead of the violatus (31.26) that is simply omitted in all its siblings, which may point at a material issue causing illegibility in F's exemplar.

At a certain point, this valuable Plut. 76.32 – which incidentally displays marginal notes by Niccolò Niccoli – seems to have aroused too much interest for its own good.Footnote 28 Somebody cut out a folium between the current f. 14 and f. 15. The verso of the lost page contained paragraphs 1–6 of the Pro Archia, which makes F acephalous (it starts with afficiebatur at 6.29). Luckily, however, F was copied before its mutilation into a second-hand addition to the present Diez. B Sant. 149 of the Berlin Staatsbibliothek (S), a manuscript that once belonged to Sozomenus of Pistoia,Footnote 29 and that beyond copying F's particular readings not only adds a few errors of its own, but also – along with only half a dozen recentiores – reads voci instead of the paradosis' voce at 19.32.

Because of the quality of B and F, the other manuscripts belonging to β1 need no further classification for the purposes of this study. While certainly not the origin of all younger members of β1, Poggio Bracciolini's copy Plut. 48.11 (M) may be singled out as a sample, as it is probably the oldest of them all (c. 1415) – although one might as well cite its close relative Plut. 48.10, which was written in 1416.Footnote 30

A PROMINENT MEMBER OF THE β FAMILY

Manuscript D 69 inf. (B) of the Biblioteca Ambrosiana in Milan, then, contains various rhetorical and philosophical Ciceroniana, in which the Pro Archia is the only oration. This miscellaneous codex has been included in a catalogue of Ambrosiana manuscripts and printed editions that somehow relate to Petrarch, because of ‘una massiccia rappresentanza di opere di filosofia morale, già predilette da Petrarca, come i Paradoxa Stoicorum, il De fato, le Tusculanae disputationes’, and because of a marginal note referring to Petrarch's De remediis utriusque fortunae.Footnote 31 The catalogue entry contends that B has no direct links to the branch of the Pro Archia transmission that stems from Petrarch, and strangely considers it ‘un rappresentante impuro della cosidetta classe Z’,Footnote 32 while B is in effect one of the soundest of all Itali, and a twin of β1, the common father of all later members of the β family. Moreover, the content of the manuscript (Somnium, Paradoxa, Tusculanae, De republica 3.33, De fato, Timaeus 3–7, Pro Archia, Topica, De officiis), in combining the Pro Archia with a fair anthology of Ciceronian philosophica, shows some affinity to several of the oldest Florentine β codices, like F (De senectute, Paradoxa, Ep. Bruti, Pro Archia, Somnium, a fragment of Pro rege Deiotaro and Seneca's De ira), and another Laurenziana manuscript, probably from the second decade of the Quattrocento: Strozzi 37 (Paradoxa, De amicitia, De senectute, Somnium, De fato, Academica, Pro Archia). Their common ground may supply a hint at the composition of the original β manuscript, and at any rate it seems safe to assume that in the original β the Pro Archia was the only oration present.Footnote 33

Yet B may be close to Petrarch for a reason other than its primitive readings: missing in the 2004 catalogue entry is any reference to the verses on the last folium (f. 143r) of the codex, mentioned none the less in the Ambrosiana's online catalogue entry of B as ‘Carmen, inc. Romanum eloquium sparsum per flumina mille.’Footnote 34 The complete ‘carmen’ (written by at least a coeval hand, if not the one that copied the preceding De officiis) goes as follows (I transcribe ad litteram):

Romanum eloquium sparsum per flumina mille
hinc fluit, he flammam multam fecere faville.
Excellunt cunctos hi libros philosophorum,
libri quos fecit tres Tullius Officiorum.
Eloquio praeceps magnus memorabilis augur
Tulius indigna cede peremptus obis.
Tertius omnipotens implevit nomine claro
ingenium ceso corpore morte caret.
Vivit et ingenti pollet cum laude per orbem,
cuius in hoc tumulo membra sepulta iacent.

To the left of the third and fourth verse, a bracket points to the word ‘Augustinus’ – as these lines (Anth. Lat. 785 Riese – with a striking productio in arsi in philosophorum) were traditionally attributed to Augustine. Not uncommonly they precede or follow manuscript witnesses of Cicero's De officiis. The following three distichs (Anth. Lat. 613 Riese) are annotated ‘Maximinus’ and are indeed part of the Epitaphia Ciceronis duodecim sapientium, where they are attributed sometimes to Maximinus, and sometimes to Asmenius.Footnote 35

The interest of the assemblage, however, is in the unidentified first two verses, whose origin has gone unnoticed: Romanum eloquium sparsum per flumina mille | hinc fluit, he flammam multam fecere faville are an example of the type of rhymed hexameters that Petrarch used to compose for inscription as an ex libris at the opening of his manuscripts. In the Biblioteca Laurenziana's Plut. 50.43 (f. 1r), the very same distich has the subscription ‘F. Petraccus de Cicerone’.Footnote 36 Taken together, the soundness of the Pro Archia's readings and the occurrence of this distich may reveal that ultimately B as a whole is much closer to Petrarch's library than has hitherto been acknowledged.

ESTABLISHING THE γ FAMILY

The existence of the β group disproves the exclusive bifurcation of a and ϕ – as the β errors are absent from both – but before moving on to a closer examination of the precise position then to be assigned to a, compared to this newly established β family and the well-known but only very partially reported ϕ family, it must also be made clear that ϕ as it is represented in modern editions is in fact an inaccurate reduction of a more complex reality, and not only because of the overlooking of the β family. Indeed, the arbitrarily selected ϕ readings in the editions, and even the consensus of Palat. lat. 1820 (P) and Vat. lat. 9305 (Q), two manuscripts that since the publication of the most recent critical editions have been acknowledged as among the more primitive members of this branch, only give a partial view of what I will call from now on the more comprehensive branch γ instead of ϕ. This branch contains at least two thirds of all our manuscripts, that is, more than 150. It divides into several – often intertwining – sub-branches, yet all uncontaminated members of this sprawling γ family display a clear series of distinctive errors:

  1. 5.22 domum ] in domum γ

  2. 5.24 fuerit ] fuit γ

  3. 6.25 Q. Metello ] Q. om. γ

  4. 13.6 adeo ] ideo γ

  5. 20.17 item ] om. γ

  6. 22.1 rebus ] om. γ

  7. 23.14 est ] om. γ

  8. 24.17 Homerum praeconem ] praeconem Homerum γ

  9. 25.28 videmus ] vidimus γ

  10. 30.12 est ] sunt γ

  11. 31.24 laudis ] laudum γ

A sound, old member of the γ family, written in Venice by Petrus Anzotus between 1396 and 1415, is Laur. Plut. 53.12 (C).Footnote 37 While not containing any of the marginal notes and corrections present in P, Q and T, it has some interesting, peculiar readings.Footnote 38 Indeed, at 7.1 it is the only manuscript that reads the correct Silvani, assigned by Clark to Hotoman and by Kasten to Manutius, against the other witnesses' Sil(l)ani, and at 31.17, along with two dozen much younger, emended codices, it has the quantum generally accepted over the paradosis' quanto.

Most members of γ3 – the subgroup of γ to which Q and T belong, that is – read ac instead of aut at 2.14, est delatus for delatus est at 11.17 and fructum gloriae instead of gloriae fructum at 23.7. Furthermore, while in the myriad of recentiores various clusters of manuscripts can be discerned that to a varied extent share the particular errors of γ1 (the consensus of γ3 and P), like the widely spread est quisquam instead of quisquam est at 20.12, there is a group of some fifty-odd other γ manuscripts – I will call it γ2 – that in every one of these instances along with C have the correct reading. Beyond the common γ errors listed above they distinguish themselves by consistently reading praeconium facile instead of facile praeconium at 20.13.Footnote 39 In the apparatus of various editions, this transposition is credited to k (Paris, BNF, lat. 7779), which is actually a late, contaminated witness. Most of these manuscripts also omit the a before forensi at 3.22, and at 5.19 for eos read eius – although in the latter case more often than not this harsh reading has subsequently been corrected into eiusmodi. In my stemma, however, I am assigning manuscripts to this γ2 subgroup on the base of the praeconium facile transposition alone, and to γ3 relying on est delatus, mostly in combination with fructum gloriae and often also with longiusculis versibus (25.29).Footnote 40 While omissions are often considered a watertight argument for proving or even rebutting dependencies, in a crowded environment such as ours transpositions turn out to be much more cogent. In Quattrocento Italy, Pro Archia copies were quickly ubiquitous, and in the event of even the most superficial collation and contamination, omissions are the first to disappear (hence also the spreading success of interpolations), disguising relationships often still indisputably demonstrable by a combination of metatheses.Footnote 41

More primitive exponents of γ2 may be singled out: Modena, Biblioteca Estense Lat. 442, which shares some of Petrarch's postille with P, Q and T;Footnote 42 Casale Monferrato, Seminario vescovile, MS' 1 c 5;Footnote 43 and Vat. lat. 1485 (R), which has in its text a few of the variae lectiones of Q too. But we will return to R later. In the meantime it may suffice as a conclusion of this assessment of γ and β that it comes as no surprise that all manuscripts Kasten based his ϕ on turn out to belong to the γ family.

THE UNIQUENESS OF a

The status of a as the only reflection of Petrarch's primitive text before he shed his brilliant philological light on it, conserved in splendid Florentine isolation, turns out to be as untenable as its lonely opposition to all other Itali. At least in the Pro Archia, Clark's cherry-picking of peculiar variants from a seems to have generated the idea that in all other instances it presents correct readings – which is absolutely not the case. Indeed, three other, much younger manuscripts – Vat. lat. 1837 (b), Ottob. lat. 1184 (c) and Marc. lat. Z428 (1810) (d)Footnote 44 – have a long list of corruptions in common with a. The first probably stems from the late 1420s, the other two from the 1450s. Along with a, this group reads:

  1. 1.2 me non] non me

  2. 1.10 aliquando] aliis

  3. 5.17 om. de

  4. 5.20 om. etiam

  5. 6.1 iure ac foedere] federe ac iure

  6. 9.31 nullam] ullam

  7. 13.31 reprehendat] reprehendit

  8. 13.1 ludorum celebrandos] celebrandos ludorum

  9. 16.1 om. et

  10. 18.29 quod] qui

  11. 19.7 expetunt] expectant

  12. 19.9 contulerit] contulit

  13. 21.24 Armeniorum] armatorum

  14. 23.13 tum] tamen

  15. 24.19 corpus eius] eius corpus

  16. 28.20 nos ] non

I have quoted a multitude of examples in order to illustrate the heavily corrupt nature of a, which should have prevented it from being smuggled so selectively into a critical apparatus without cautioning about its real value.Footnote 45 The number of mistakes in a is quite appalling, and the list above could easily be tripled if one were to quote all a errors shared by any two out of b, c and d. The precise relationship of these three younger manuscripts to each other and to a itself is not easy to establish: at face value, they seem to stem from a sibling or from the very source of a, as they all lack several of a's omissions, yet collation shows all three to bear unmistakable signs of compilation – which accounts for the fact that only very few of the more absurd particular errors of a are shared by all four. Indeed, at 30.12, b has γ's abfutura sunt and the cum statuas that from β spread over a fair number of γ manuscripts, while at 5.22 c and d have γ's in domum, and at 13.6 d also has γ's ideo for adeo. Beyond these instances of contamination, in numerous passages varying pairs of members of this group distinguish themselves from the remaining two. The only place that sees b, c and d united against a's correct reading is their calamitatis for calamitas at 9.27, but this error is not uncommon in various other, independent recentiores, and therefore it may have been introduced (or even have occurred) in all three independently. So in the end, b, c, and d may be descendants of a itself that were partly corrected after collation, although the hypothesis that they share a common source with a may still be the safer one.Footnote 46

THE α FAMILY

Apart from rebutting the assumption that a either left no offspring or had no direct siblings, the issue is not particularly relevant, as a's source in addition generated a few more distant (and certainly less corrupt) relatives. This lost sibling of a (which I will call α2) was the exemplar of both Ottob. lat. 1463 (O) and Plut. 48.25 (N). These twins were probably copied in the San Gimignano area in, or even before, the 1420s, clearly by the same copyist, and they have in common enough variations to justify the existence of a common origin from which α2 and a derive. In the following cases, O and N share the same error against the correct reading in the other witnesses, a included:

  1. 1.9 ingrediendam] gubernandam O N

  2. 4.10 sic] sicut O N

  3. 6.2 ipse] om. O N

  4. 7.6; 7.8 praetorem] praetores O N

  5. 10.7 utitur] om. OacN

  6. 10.8 se] om. O N

  7. 14.20 animum] animam O N

  8. 18.19 quoniam] qui O N

  9. 22.5 legibus] de legibus O N

  10. 26.5 tamen] mihi O N

  11. 29.30 dimicaret] dubitaret O Nac

  12. 30.13 certe] om. O N

  13. 32.28 simpliciterque] suppliciterque O N

Other variants unite a (and its three followers, with the exception of a few instances where b, c or d may be bearing the traces of contamination) and α2 against the other witnesses:

  1. 1.3 studiis] studio a b c O N

  2. 5.16 ceterisque praemiis] premiis ceterisque a b c O N

  3. 5.22 etiam tum] et a b c d O N

  4. 9.24 solae] sola a b c d O N

  5. 16.5 ut] om. a b c d O N

  6. 21.24 Cyzicenorum] Agricenorum a b c d O N

  7. 22.3 nostrum laude] laude nostrum N laude nostrorum O voce nostrum a d voce c

  8. 31.18 eius modi] est huiusmodi a b c d O N

  9. 31.23 qui] et qui a b c d O N

Another important manuscript hitherto ignored is Avignon, Bibliothèque Municipale 1215 (A).Footnote 47 This fifteenth-century collection of speeches is to α what B is to β – that is, the best witness of its branch. It displays the correct reading in the preceding two lists, yet has a few errors in common with α1 – that is, the common source of α2 and a – that prove their common origin:

  1. 11.11 primis] ApcBal in primis Aaca b c d O N

  2. 15.29 ego] om. A a b c d O NBal

  3. 19.8 vivum] virum A a b c d O NBal

  4. 26.6 est] om. A a b c O NBal

  5. 26.10 despiciunt praedicari] disputant et praedicari a c d Nac despiciunt add. s.l. N despiciunt et praedicari O dispiciunt et praedicari b disputant praedicari ABal

  6. 31.23 domesticis periculis] periculis β1 periculis domesticis A a b c d O NBalFootnote 48

Furthermore, A is the only manuscript not belonging to γ that presents marginalia that seem to go back to Petrarch. Indeed, apart from several explanatory notes that in γ are absent or differ,Footnote 49 there is one striking correspondence: at 18.27, the marginal note in both A and Q reads ‘Ars ceteros studiosos, natura poetam facit’.Footnote 50 Moreover, A contains variae lectiones witnessing Petrarch's philological reflection. At 21.22, A reads naturae regione (with G and E), but in the inner margin the copyist has added the alternative reading ‘aliter: natura’ (al' nã). Most β and γ manuscripts indeed read natura regionis. At this point, however, Q and P read naturae regione too, and have no marginal note – but T has incorporated the double reading, as its text reads: natura (probably corrupted from naturae) regione vel natura regionis. At 25.28, A has videmus, like the other members of α, with G and E, against the vidimus of β and γ. The interlinear l' vidimus in A, however, indicates that here too Petrarch had a double reading, and vidimus must not be considered a Bindefehler of β and γ.Footnote 51 In the same paragraph, the first hand al' iussit alongside iubere in A explains the iussit of all α1 manuscripts: apparently its copyist went with the alternative reading, a Schlimmbesserung that subsequently met with some success in various contaminated younger manuscripts belonging to γ.

At 15.30 the great majority of our manuscripts read the correct ratio, yet α's O and N and β's B have oratio. This is no trivial coincidence of a common confusion: Q reads oratio with the marginal alternative vel ratio, and A has oratio al' ratio – for once incorporated into the text, just as in T and P at this point – which seems enough evidence for accepting the presence of a double reading at the top of all branches.

Although the number of members of the α family is limited to roughly a dozen,Footnote 52 it can be clearly distinguished from β and γ. The unique status of a in the Pro Archia is in other words untenable because α has siblings and because there are Petrarchan variants in α, as proved by A.Footnote 53 Probably such variae lectiones or double readings were not entirely absent from β either. At 27.16, the manuscripts are divided between locati and the convincing emendation togati. While α has locati, and γtogati or a double reading, the distribution in β seems to challenge the stemmatic reconstruction offered above: B and M have locati, but F reads togati. Unless we prefer to suppose that the shrewd Latinist behind F's various sophisticated innovations independently reinvented Petrarch's emendation, the more likely explanation seems to be that β had the same double reading as γ, and that in the transmission of α either the copyists went with the original reading, omitting the emendation, or the emendation was absent.Footnote 54

Of all three branches that I have distinguished so far, γ alone bears traces of an original packaging of Lapo's quartet with the Pro Archia. Since in the various primitive members of both α and β the Pro Archia is not systematically preceded or followed by a specific oration (or series of orations), we may assume that in these two branches the speech initially travelled alone and that only in later stages was it incorporated into more comprehensive collections. Clark's theory that the unnamed other speech discovered by Petrarch in Liège was Pridie – an assumption based on no more than the order of the speeches in ‘Lapo's’ a – can therefore be dismissed as pure conjecture. There are indeed no compelling reasons to assume that the order of the various components in a reflects Lapo's personal intervention or the order or packaging in pre-existing nuclei in his library. Furthermore – if the order of the speeches is supposed to be a decisive indication – the insertion of Pro Archia, Pridie and Red. sen. between the first and the remaining three members of the quartet is as much an argument for as against the coupling of Pro Archia with Pridie in Petrarch's present. Ultimately, as Petrarch in his letter from Parma to Lapo mentions the sending of one speech only, the most obvious conclusion seems that for whatever reason – most likely because it was a well-known text already circulating in Italy – he felt no need to mention the other Liège discovery or send it as a bonus.

A FOURTH BRANCH

A last, most interesting manuscript is codex 59 of the Biblioteca Guarneriana in San Daniele del Friuli (D). This relatively late witness was copied in February 1439 in Lavariano (close to Udine and some 20 miles south of San Daniele) by Niccolò da San Vito, as one of the first codices for the library of Guarnerio d'Artegna.Footnote 55 Its composition is striking, both in its own right because of the specific order of the texts, and because of the conspicuous parallels to the already mentioned and equally northern Italian Vat. Pal. Lat. 1485 (R). Indeed, the content of D is largely identical to that of R, but the order of the speeches is slightly different.Footnote 56 Yet this difference reveals an interesting fact: D is the only known witness presenting the three Caesarianae (albeit mixed up among themselves, as Deiot. and Lig. traded places) followed by Pro Archia and then Lapo's quartet – oddly reversing the ‘canonical’ order in numerous members of the γ branch, and most likely in γ itself. Yet it seems highly improbable that a careless copyist who mixes up the order of a series of speeches would do so with apparent respect for the original nuclei in the collection, that is, moving units rather than single speeches, if no material division were present in his exemplar; such a practice would indeed seem unlikely if he were just copying a pre-existing collection of speeches without perceptible internal subdivisions.

Furthermore, collation shows that D, while displaying a series of particular errors that are absent from all three other branches, shares no errors whatsoever with γ or with any of its sub-branches – whereas, it should be remembered, its ‘mirror manuscript’ R undeniably belongs to γ – nor exhibits any of the typical errors of α or β. In the absence of any traces of a crafty compilation that wiped out all specific variants of the other three families, D must therefore by itself constitute a fourth branch δ, stemming independently from Petrarch's π.Footnote 57 The compositional similarity of D and R can be explained if one assumes that δ's exemplar in the Pro Archia, that is π, was up to a certain moment being kept in close proximity to γ2 itself, in an environment where the speeches were available as physically separated (or at least clearly distinguished) entities, thus allowing for the slightly altered copying order in R and D.Footnote 58 Only one place, in my opinion, fits this description of such circumstances: Petrarch's own library, or at least a collection of manuscripts that had belonged to that library and that for some time stayed together. Making use of this collection of separate libelli, the compiler of δ must have taken up first the Caesarians and then transcribed the separate Pro Archia manuscript π. When he copied the quire containing Lapo's quartet, he may have used a model that contained the full quintet, as it has been incorporated in γ, and realized after having transcribed the first four speeches that he had already dealt with the final text contained in this quintet, and so omitted it to prevent duplication. Another possible explanation is that he had access to a more primitive copy of Lapo's quartet that still existed as a separate codicological unit and that was independently used as the source text for the compilation of the quintet.

This reconstruction allows for the similarity between R and D, and also explains a few striking resemblances between δ and one of the other branches that at first sight may seem to conflict with a quadripartite stemma. In fact, D reads ratio (15.30), videmus (25.28) and iussit (25.30) with α1, but in these three instances the presence of these readings accompanied by their alternatives oratio, vidimus and iubere in A proves the presence of double readings in π, which is corroborated by β's split between oratio (B) and ratio (β1). As copyists, when confronted with double readings, clearly oscillated between copying both alternatives, making their own choice and (most likely) simply omitting or overlooking the marginal variant, these double readings, before sooner or later disappearing, generated some transverse textual agreements that only when taken at face value challenge the proposed stemma. It would therefore be excessive to conclude from the addition of the first sive (30.12) in both β and γ that these two branches are to be reduced to one, or that α and δ were copied before Petrarch added this sive to his π, and β and γ only afterwards. For it is not impossible that this alternative reading was omitted by α and δ (the latter bearing no traces of any double readings whatsoever). In another instance too, A proves very useful: at 12.26 se ita unites β, γ and δ against the ita se that puts a in line with G, E and V, but as A has se ita just like B, C and D, se ita turns out to be an innovation of α1 that by mere coincidence joins G, E and V.

CONCLUSION

We can safely assume, I believe, that the Pro Archia as transmitted by α, β and δ goes back to Petrarch's original transcription from Liège, which contained only a few marginal annotations and emendations. At least one copy of this version of the text was sent out during his lifetime, to Lapo, while Petrarch may have made and kept for himself a personal copy, γ, filled with the more lavish critical and other annotations that we find all over the γ branch. This reconstruction of the transmission into four branches leads to the conclusion that all fifteenth-century copies of the Pro Archia do indeed go back to Petrarch's text. All witnesses stem from π, against G E V, and not a single one displays superior true readings that might suggest traces of a branch descending from another copy of the Liège manuscript, independently of Petrarch.

Only after Petrarch's death, then, did γ create a much larger offspring than its former possessor had ever permitted during his lifetime. We know that a part of Petrarch's library stayed in Padua with his son-in-law and heir Francescuolo da Brossano, and that P was copied in Padua in 1394.Footnote 59 The subscription about the quinque orationes in P and T and the physical fact of T indeed being limited to this package points to the existence of a libellus containing Lapo's quartet with the addition of the Pro Archia. More than once the question has been raised whether the numerous late fourteenth- and fifteenth-century manuscripts that contain a fixed series of eleven Ciceronian speeches – that is Lapo's quartet, Pro Archia, the three Caesarians, Pro Quinctio and Pro Flacco – reflect a collection made by Petrarch himself or one realized after his death. Most scholars, if tentatively, seem to prefer to ascribe the process of collection to Petrarch himself.Footnote 60 In my opinion, the fact that the original quintet codex T is a twin of Q rather than siding with P, and the reversing of the order of the nuclei in D rather tilt the balance towards surmising that the series of eleven speeches was an assemblage produced only in a subsequent phase, while the composition and stemmatic position of D also indicate that the Paduan part of the heritage must have contained if not π itself, at least a copy of it that was independent of γ. As Francescuolo received only duplicates, π may have been part of the other, bigger part of the heritage, which in 1388 moved from Padua to the Visconti library in Pavia. It is tempting to see in B an early result of the arrival in that part of Italy of Petrarch's rich library.Footnote 61

While γ was most likely a Petrarch transcript – or at least compilation – of the quintet (witness the subscription in P and T), it is perfectly possible that γ itself was copied into more than the three descendants displayed by the stemma below. As the main concern of my research has been to offer a documented alternative to the untenable bipartition between a and ϕ, sifting primitive from more sophisticated later manuscripts, I have focussed on establishing the other families rather than on unravelling the prolific γ branch, although such an effort may still be of interest for the reconstruction of the transmission of other orations. In their transmissions, D may very well align with γ (as in my opinion both assemblages have a direct link to Petrarch's library), while the Pro Archia's solitary tradition in the most primitive members of α and β makes it more likely than not that transmission lines in these two branches diverge from what has been outlined above.Footnote 62

In any case, the more likely candidate to have been Petrarch's gift to Lapo is α, given its earlier proliferation in Florentine codices that furthermore contain members with no traces of the quartet or quintet coupling. This hypothesis also fits in with the fact that when in 1379 Coluccio Salutati wrote to Lombardo della Seta asking for transcripts of Ciceronian texts from Petrarch's library, he asked for all four members of the quartet, but not for the Pro Archia – he already had it, from the separate copy that Petrarch himself had sent to Lapo.Footnote 63 Furthermore it has been argued – in support of the theory that a was Lapo's own copy – that when Lapo in 1378 was exiled from Florence and his possessions were sold, he also lost his books, which is supposed to explain why at his death in 1381, the inventory of his library redacted by his sons bears no trace of Ciceronian speeches.Footnote 64 Of course this does not disprove previous ownership of a by Lapo, nor the fact that a, although not Lapo's autograph copy of the antique quartet libellus, can still be a descendant of it. Yet the fact that Salutati barely one year after Lapo's exile from Florence writes to Padua asking for, among others, the four speeches that Lapo certainly once owned – and intriguingly does so mentioning them all four one after another – suggests that if Lapo did not take the libellus with him, the auction of it had at least made it unavailable to Salutati. It is therefore even harder to explain why this very manuscript a, Salutati's late Ciceronian compilation, which in its last parts probably was copied not before the 1380s (and the manuscript indeed concludes with Pro Milone, Pro Plancio and finally Pro Sulla) and which consequently was written after Lapo's banishment from Florence and after Coluccio's explicit request to Lombardo della Seta for the orations that had been in Lapo's possession, is allegedly the only direct descendant of Lapo's quartet (although in a the four orations are not present as such, as an uninterrupted series that is) without the detour by way of Petrarch.Footnote 65 The only possible conclusion, for the time being, seems to be that at any rate Salutati did not obtain the quartet from Lombardo or from Petrarch's estate in Padua, but that he did succeed in gaining access to a descendent of Lapo's codex.

The question remains, nevertheless, how likely it is that all Italian manuscripts of the quartet except a stem from the copy that Petrarch made from the text he received from Lapo. When the witnesses to these texts number well over 200, it seems editors have been overconfident in relying on so narrow a sample. How can one be so sure that no other manuscripts survive that stem directly from Lapo's codex, or even from the unknown source that he took his text from, or that twins of his codex or its source left no offspring? If these questions have ever been raised, they have not been adequately answered. For the reasons set out above, D may be a good start for improving on the reconstruction of Petrarch's copy of the quartet, and of Petrarch's version of other orations as well, beyond the overrated P and Q.Footnote 66 Older members of α and β that contain the quartet not immediately followed by the Pro Archia seem to offer the most promise of yielding traces of Lapo's text without the detour by way of Petrarch, or at least for leading us to a stage in the transmission that precedes the compilation of the quintet, as quite a few manuscripts do contain Lapo's quartet as an uninterrupted unit, but not immediately followed by the Pro Archia.Footnote 67

Meanwhile, in the Pro Archia four independent families can be seen branching from Petrarch's text, each of them represented by important manuscripts emerging from neglect. Their consensus allows a much firmer reconstruction of the lost Leodiensis and a better appraisal of its relation to the other three primitive witnesses to this oration.

APPENDIX: THE MANUSCRIPTS OF CICERO'S PRO ARCHIA

In this list the manuscripts and folia of the speech are followed by an (approximate) dating, usually based on the existing literature, mostly the relevant library catalogues; then (if applicable) the corresponding entry in Silvia Rizzo's Catalogo dei codici della Pro Cluentio ciceroniana (Genoa, 1983) and the corresponding manuscript group in Elżbieta Olechowska's edition of Pro Cn. Plancio, Pro C. Rabirio Postumo (Leipzig, 1981). As in numerous members of the γ branch the Pro Archia (A) is preceded by the Pro Sulla (S), the last speech of Lapo's quartet, and followed by the Pro Marcello (M), the first of the Caesarians, I have entered a note wherever this order is present.

The ₪ symbol indicates manuscripts not seen; most of them are recentiores, and quite a few display the SAM sequence, which makes their belonging to the γ branch very likely. The last column has the sigla used in the discussion and stemma above.

In some cases I have preferred to suspend judgment, or rather to refrain from allocating manuscripts that I did see to a specific branch. Apart from some instances where the partially transmitted text is too short to allow a decisive classification (or where I only saw a partial reproduction), these are corrupt or heavily contaminated manuscripts – and mostly both. Usually I could have surmised an affiliation to β or γ, but as a major purpose of this list is to serve future manuscript research on other Ciceronian speeches, it seemed more sensible not to simplify an often complex stratification by explicitly classifying witnesses that probably hold no promise for stemmatic purposes anyhow.

Likewise, the attributions to γ2 and γ3 are not exclusive: the absence of classification into a subbranch of γ should not be interpreted an implicit negation of a possible adherence.

For easy reference, and for the sake of completeness, I have included the four manuscripts that do not derive from Petrarch's copy (nos. 7, 17, 177 and 229), and I have retained some others that were lost during wartime (28, 81 and 197) but have been described in detail. I did leave out Rizzo's lost codices viii (Escorial, SAM) and x (Reims), in the absence of more detailed information.

Several manuscripts present an incomplete text: nos. 6 (jumps from §15 to 19); 86 (jumps from §8 to 25); 163 (starts §13); 180 (§§1–6); 181 (jumps from §4 to 15); 186 (§§1–16); 187 (excerpts); 189 (§§1–4); 207 (fragment from §18); 211 (quotes from §§1–5, 14, 15, 16–20, 28–29, 31); 242 (§§1–17); 249 (the quire containing the text of Pro Archia, once starting on f. 209r, as can be deduced from the catchwords at the bottom of f. 208v, is now missing); 267 (§§1–19) and 268 (§§1–25).

Footnotes

*

I am very grateful to Michael D. Reeve, who most generously shared his unfailing Ciceronian knowledge with me, to Monica Berté, trusted source of various consulenze petrarchesche, and to Tom Deneire, fellow-traveller along Archian roads, for their helpful comments on drafts of this article. I am also much obliged to Jan Papy for his enthusiastic support of this project, and to Stephen Oakley for his thorough review and valuable suggestions.

References

1 For an assessment of the importance of this discovery, ‘one of the earliest in a succession of discoveries that by the 1570s had transformed the canon of classical Latin literature’, see Reeve, M.D., ‘Classical scholarship’, in Kraye, J. (ed.), The Cambridge Companion to Renaissance Humanism (Cambridge, 1996), 20–46CrossRefGoogle Scholar, at 20–6.

2 Circa quintum et vigesimum vitae annum inter Belgas Helvetiosque festinans, cum Leodium pervenissem, audito quod esset ibi bona copia librorum, substiti comitesque detinui, donec unam Ciceronis orationem manu amici, alteram mea manu scripsi, quam postea per Italiam effudi; et ut rideas, in tam bona civitate barbarica atramenti aliquid, et id croco simillimum, reperire magnus labor fuit (Sen. 16.1). Petrarch's links with the Low Countries are discussed by Dykmans, M., ‘Les premiers rapports de Petrarque avec les Pays-Bas’, BIBR 20 (1939), 51–122Google Scholar, at 51–63. An overview of the more recent publications about Petrarch and the Low Countries is given by Tournoy, G., ‘La fortuna del Petrarca nei Paesi Bassi’, in Blanc, P. (ed.), Dynamique d'une expansion culturelle: Pétrarque en Europe XIVe–XXe siècle: actes du XXVIe congrès international du CEFI, Turin et Chambéry, 11–15 décembre 1995 (Paris, 2001), 583–94Google Scholar, at 584 n. 6.

3 See for an analysis of the Collatio laureationis among others Verhulst, S., ‘Le Pro Archia comme paradigme d'amplification: réflexions sur le discours épidictique de Pétrarque à la Raccolta aragonese’, in Galand-Hallyn, P. and Hallyn, F. (eds.), Poétiques de la Renaissance. Le modèle italien, le monde franco-bourguignon et leur héritage en France au XVIe siècle (Geneva, 2001), 346–60Google Scholar; and more recently Mazzotta, G., ‘Petrarca e il discorso di Roma’, in Finucci, V. (ed.), Petrarca: canoni, esemplarità (Rome, 2006), 259–72Google Scholar.

4 Collatio laureationis, 1.6–7: Quanta, inquam, sit naturaliter difficultas propositi mei ex hoc apparet quod, cum in ceteribus artibus studio et labore possit ad terminum perveniri, in arte poetica secus est, in qua nil agitur sine interna quadam et divinitus in animum vatis infusa vi. Non michi, sed Ciceroni credite, qui, in oratione pro Aulo Licinio Archia, de poetis loquens verbis talibus utitur: ‘Ab eruditissimis viris atque doctissimis sic accepimus: ceterarum rerum studia et ingenio et doctrina et arte constare, poetam natura ipsa valere et mentis viribus excitari et quasi divino quodam spiritu afflari ut non inmerito noster ille Hennius, suo quodam iure, “sanctos” appellat “poetas”, quod deorum munere nobis commendati esse videantur’; hec Cicero. See further Godi, C., ‘La «Collatio laureationis» del Petrarca nelle due redazioni’, Studi petrarcheschi ns 5 (1988), 1–58Google Scholar, at 30–1. Petrarch is quoting Arch. 18.

5 The exchange has been reconstructed by Foresti, A., ‘Le lettere a Lapo da Castiglionchio e il suo libro ciceroniano’, in Benvenuti, A. Tissoni (ed.), Aneddoti della vita di Francesco Petrarca (Padua, 1977), 242–50Google Scholar.

6 Var. 45: Orationem Tullianam pro Licinio Archia, quam pollicitus sum tibi, praesentem mitto refertam miris poetarum laudibus. Iuvabit puto fide digno teste cognoscere, quod studiis quibus delectamur praeco ingens et praeclarissimus Orator accesserit, cuius rei admonuisse te velim, ut rem licet parvam in pretio habeas. Neve pro una tres eiusdem oratoris remisisse te pigeat, quas fateor ipse mecum detulissem, et sicut uno, sic duobus te libris uno tempore spoliassem per fiduciam, nisi propere Philippicas dimisissem, neque id eo proposito ut eis perpetuo careres, sed ut accuratius et littera floridiore transcriptas ad te quamprimum remitterem. Ego enim usque adeo vetustati oculos assuefeci ut novam scripturam qualemcunque fastidiam. Super hoc nichil amplius. Tuus sum et Lapum meum lappis tenacissimis amoris obsitum fuisse nunc sentio, a quo divelli nequeo. Sic te semel complexus animo inhaereo. Feliciter vale. Franciscus Petracch. Tuus. Parmae, octavo Idus Ianuarii festinanter valde. See Pancheri, A. (ed.), Lettere disperse. Varie e miscellanee (Parma, 1994), 102–5Google Scholar.

7 Fam. 7.16: Milonianam Ciceronis cum reliquis accepi; gratias ago. Non michi nunc primum tue mentis indulgentia nota est; rescribi faciam, et remittam. Cf. Fam. 12.8, a letter ‘about Cicero and his works’: Quod autem nominatim ad libellum tuum attinet, amice, et Milo defensus et Laterensis offensus et Silla excusatus et Pompeius laudatus aderant.

8 Fam. 13.6: Quam non apposui quoniam orationem illam, ab extremis olim Germaniae advectam, dum loca illa visendi ardore iuveniliter peragrarem, et anno altero in patriam vobis optantibus transmissam, habetis studioseque legitis, quod in literis inde venientibus recognosco.

9 ‘Puto igitur hunc librum ad ipsum Lapum redire et vel ab ipso vel in eius usum fuisse transcriptum.’ Clark, A.C. (ed.), M. Tulli Ciceronis Orationes. Pro Tullio. Pro Fonteio. Pro Sulla. Pro Archia. Pro Plancio. Pro Scauro (Oxford, 1911)Google Scholar, ix.

10 Preceding a series of Cicero's Orationes, this manuscript also contains the Tusculanae, De senectute and the Paradoxa Stoicorum. The most extensive description has recently been printed by De Robertis, T., ‘Un'edizione di Cicerone in gran parte copiata da Salutati (Firenze, Biblioteca Medicea Laurenziana, Pluteo 23 sin. 3)’, in De Robertis, T., Tanturli, G. and Zamponi, S. (edd.), Coluccio Salutati e l'invenzione dell'Umanesimo (Florence, 2008), 321–3Google Scholar, no. 104.

11 See Billanovich, G., Petrarca e il primo umanesimo (Padova, 1996), 97–116Google Scholar (a reprint with some additions of id., ‘Petrarca e Cicerone’, in Miscellanea Giovanni Mercati IV, Letteratura classica e umanistica [Vatican City, 1946], 88–106), esp. 109 n. 40: ‘Dunque il Petrarca, oltra la Pro Archia, dovette comunicare pure questa orazione agli amici di Firenze. Ma è un nuovo indice di percezione geniale che egli non abbia amato né unirla alla Pro Archia, né richiamarne delle citazioni.’ This is not the only instance where Billanovich is cutting corners, founding his speculation on argumenta ex silentio.

12 Rizzo, S., ‘Da Firenze alla biblioteca del Petrarca. Scambio di doni ciceroniani fra Petrarca e Lapo da Castiglionchio. Biblioteca Medicea Laurenziana, XXIII sin. 3’, in Feo, M. (ed.), Codici latini del Petrarca nelle biblioteche fiorentine (Florence, 1991), 9–14Google Scholar, at 12–13: ‘L'interesse di a è dato proprio dal suo essere immune dagli interventi petrarcheschi, così da costituire quasi una cartina di tornasole che ci aiuta a distinguere la lezione originaria dalle congetture del Petrarca. Ciò ha permesso di stabilire che lezioni registrate dagli editori negli apparati (e talvolta nel testo) come provenienti dalla familia Gallica [that is, the branch containing all other descendants of Petrarch's copy] sono in realtà congetture, e a volte semplici glosse penetrate nel testo, risalenti al Petrarca.’ In these pages, Silvia Rizzo gives the best overview of the whole issue, recapitulating the contributions of Clark, Billanovich and others. In the same year, Albinia de la Mare declared a an autograph by Coluccio Salutati, but in the meantime, it has been demonstrated that there are several hands at work, only one of them Salutati's: see De Robertis (n. 10). A new discussion of the attribution – based only on paleographical grounds – of a to Salutati's hand, is given by Berté, M., ‘Petrarca, Salutati e le orazioni di Cicerone’, in De Paolis, P. (ed.), Manoscritti e lettori di Cicerone tra Medioevo ed Umanesimo (Cassino, 2012), 21–52Google Scholar. I wish to thank the author for allowing me to see her work before publication.

13 Rouse, R.H. and Reeve, M.D., ‘Cicero. Speeches’, in Reynolds, L.D. (ed.), Texts and Transmission: A Survey of the Latin Classics (Oxford, 1983), 54–98Google Scholar, at 86 n. 186.

14 Kasten, H. (ed.), M. Tulli Ciceronis scripta quae manserunt omnia. Fasc 19. Oratio pro P. Sulla. Oratio pro Archia poeta (Leipzig, 1966 3), ix–xiGoogle Scholar. See also Maslowski, T. and Rouse, R.H., ‘Twelfth-century extracts from Cicero's “Pro Archia” and “Pro Cluentio” in Paris, B.N., MS lat. 18104’, IMU 22 (1979), 97–122Google Scholar, at 99–102, for a synthesis in English.

15 Rouse and Reeve (n. 13), 86 n. 187: ‘So H. Nohl (Vienna – Prague – Leipzig, 1889), xi–xii. Clark misguidedly separated the manuscripts into a German and a French family without offering a stemma. The stemma in the Teubner of P. Reis (19331, 19492) reflects Clark's view of the tradition. H. Kasten's new Teubner … is a great improvement both in assessment of the manuscripts and in accuracy, but of the two readings that he regards as Bindefehler of G (and ω), one, the gloss Archia in 1, is not in both branches of (ω), and the other, 16 profugium for perfugium, is trivial. F. Gaffiot in vol. xii of the Budé Discours (1938) contributes nothing on the manuscripts.’ Gaffiot indeed goes to great lengths to defend even the harshest particular readings of G against sounder alternatives in other branches of the tradition, in his all-out allegiance to G's codex optimus status, already advocated by É. Thomas, Discours de Cicéron pour le poète Archias. Publié d'après les travaux les plus récents avec une nouvelle collation du Gemblacensis, un commentaire critique et explicatif, une introduction et un index (Paris, 1883)Google Scholar. Thomas claims to have examined all Paris manuscripts of the text (he enumerates 19 of the 26 that I list in the appendix below), but confines himself to classifying them all in the same group and then dismisses them as ‘des copies inférieures du texte que nous avons sous une forme plus pure dans le Gemblacensis, et qui sont à rejeter’ (at 18). I will dwell on the stemmatical relation between Petrarch's Liège find and the other major witnesses in another publication.

16 See Billanovich (n. 11).

17 Rizzo, S., ‘Apparati Ciceroniani e congetture del Petrarca’, RFIC 103 (1975), 5–15Google Scholar.

18 See also Fiorilla, M., Marginalia figurati nei codici di Petrarca (Florence, 2005), 31–3Google Scholar.

19 Ornato, E., ‘La Redécouverte des discours de Cicéron en Italie et en France à la fin du XIVe siècle et au début du XVe siècle’, in Schoeck, R.J. (ed.), Acta Conventus Neo-Latini Bononiensis. Proceedings of the Fourth International Congress of Neo-Latin Studies. Bologna 26 August to 1 September 1979 (Binghamton, NY, 1985), 564–76Google Scholar, at 572 n. 13.

20 Q breaks off in the middle of Red. pop., but M.D. Reeve argued that the manuscript may originally have contained the same parts of the two Post reditum speeches as Petrarch's well-known Ciceronian collection in Troyes, Bibliothèque Municipale 552 (the so-called Trecensis, containing most of Cicero's rhetorica and philosophica, and several speeches; see for its content P. de Nolhac, Pétrarque et l'humanisme [Paris, 19652], 1.227–30), and that the missing parts were supplemented later on in Q's apograph Paris BNF Lat. 7778. See Reeve, M.D., ‘Recovering annotations from Petrarch’, in Quaderni Petrarcheschi 9–10 (1992–3), 333–48Google Scholar, at 343 and id. (ed.), M. Tullii Ciceronis Oratio pro P. Quinctio (Stuttgart and Leipzig, 1992), xii–xvGoogle Scholar.

21 Rouse and Reeve (n. 13), at 54–7 and passim.

22 In theory some of these numerous descendants could go back to the lost Liège manuscript independently of Petrarch's copy, if the errors they all share against the united reading of G, E and V (qua for quae at 11.18; the omission of hoc at 15.28; atque instead of et at 15.29; the omission of at at 16.7) were all of them already present in Petrarch's source and he made a faithful transcription without adding any mistakes. However, no elements in either the date or the geographical origin of the more antique or primitive members of the branch point to such a hypothesis; the oldest copies stem from the last quarter of the fourteenth century, and they all are of Italian origin.

23 Gaffiot, F. and Boulanger, A. (ed.), Cicéron. Discours. Tome XII. Pour le poète Archias, Pour L. Flaccus (Paris, 1938), 33 and 23Google Scholar.

24 All textual references are to the paragraphs and line numbers of Helmut Kasten's Teubner edition (n. 14).

25 Particular errors of B, on the other hand, are adsint (8.13), praediti (10.5), om. Chii suum (19.3), statis (25.30) and expeditum (31.17).

26 As F ended up in the public library of San Marco before it was incorporated into the Laurenziana, it is not unlikely that it may have been a source for later erudite compilers, although it is of course not unlikely that its felicitous emendations surfaced elsewhere independently.

27 The only other manuscript where I have seen this reading is Bologna, Universitaria 2488.

28 Its value has been unduly downplayed by Courtney, E., ‘Notes on Ciceronian manuscripts and textual criticism’, in BICS 10 (1963), 13–16Google Scholar. Courtney lists a few interesting readings (adding ‘I should perhaps make it plain that while I regard all this as of some interest, I do not consider it of high significance’), yet before his dismissive conclusion, he had noted that the manuscript ‘is of the same general character as a, i.e. intermediate between the German family and the distinctively Italian’. The implications for a's alleged unique status seem to have escaped scholars. I am very grateful to Gabriella Pomaro, who identified Niccoli's hand while kindly examining this manuscript and sharing her expert advice with me.

29 See Winter, U., Das Handschriften-Verzeichnis der Deutschen Staatsbibliothek zu Berlin. Neue Folgen. Erster Band. Die europäischen Handschriften der Bibliothek Diez (Leipzig, 1986), 127Google Scholar.

30 See n. 54 below for the relation between Plut. 48.11 and Plut. 48.10.

31 Gavinelli, S., ‘Miscellanea ciceroniana di opere retoriche e filosofiche: la formazione dell'umanista. Milano, Biblioteca Ambrosiana, D 69 inf.’, in Ballarini, M., Frasso, G. and Monti, C.M. (edd.), Francesco Petrarca. Manoscritti e libri a stampa della Biblioteca Ambrosiana (Milan, 2004), 108Google Scholar.

32 The ‘cosidetta classe Z’ is a somewhat ambiguous reference: Rizzo (n. 17), 8 used it for referring to ‘(sigla usata dal Kasten) il perduto codice di Lapo’ (while Kasten used Z only in his stemma of the Pro Sulla), but as Gavinelli explicitly denies B a Petrarchan origin, she seems to be referring to Maslowski and Rouse ([n. 14], 101), who used Z to refer to the common ancestor of E and V – which would make B only the third member of this important family, and therefore all the more relevant.

33 In M Lapo's quartet is present as such, as it is in Plut. 48.10, but in neither case is it immediately followed by Arch. In other members of β1 the dissipation is even more substantial, which points to the independent origin of many compilations, and corroborates the hypothesis that in β the Arch. did not come with a comprehensive series of speeches.

34 http://ambrosiana.comperio.it/opac/detail/ambro:catalog:33602 (accessed 12 April 2012).

35 See the critical edition of these verses in Friedrich, A., Das Symposium der XII Sapientes: Kommentar und Verfasserfrage (Berlin and New York, 2002)CrossRefGoogle Scholar, 65 (sap. 119) and 226–7.

36 S. Rizzo, ‘I distici rimati composti dal Petrarca per i libri della sua biblioteca’, in Feo (n. 12), 97–8.

37 The manuscript has been described by Leonardi, C., ‘I codici di Marziano Capella’, Aevum 34 (1960), 1–99Google Scholar, at 41–3. Anzotus subscribed and dated several parts of the codex, which was written in several stages: as Arch. is on ff. 47r–48v, it can be surmised to have been written between 1407 (date on f. 37v) and 1415 (f. 52r).

38 While C has various second-hand corrections, some of them in rasura, the original reading can often be recovered from its Vatican apograph Rossi 1034, which reflects C's readings ante correctionem.

39 Many members of this group share, in variable constellations, several other errors, but most of them are recentiores and show signs of correction – to which this metathesis proved immune.

40 This last metathesis may, as Michael Reeve kindly points out to me, originate from Paris BNF 7778, probably the most primitive member of the group of manuscripts containing it; other members of this group – this list makes no claims for completeness – are, with reference to the numeration in the appendix: 35, 46, 47, 78, 80, 117, 119, 132, 137, 138, 143, 145, 147, 148, 152, 155, 182, 193, 195, 196, 218, 220, 233, 238, 243, 245, 246, 255. Nevertheless, the fact that many of these do not share the omission of hunc at 8.15 and some other corruptions present in BNF 7778, points to contamination in this branch, which makes it hard to establish with certainty whether all these witnesses do indeed stem from BNF 7778.

41 I have seen not a single case of a collator correcting a transposition. Indeed, a–b or ” ' transpositions tend to be introduced by copyists who are at once correcting themselves. This observation of course applies to prose texts alone, as in poetry some metatheses give themselves away more easily, being at odds with the metre, and they will often be corrected by shrewd copyists or readers.

42 See Rizzo, S., ‘Congetture di Battista Egnazio ad Orazioni di Cicerone’, in Avesani, R., Billanovich, G., Ferrari, M. and Pozzi, G. (edd.), Miscellanea Augusto Campana II (Padua, 1981), 671–80Google Scholar, at 671–4.

43 Often misquoted as 2 c 5. I am very grateful to Luigi Silvano for his photographs (and his tenacious pursuit) of the two elusive Casale manuscripts.

44 I am using these sigla merely for the convenience of this particular discussion.

45 Clark (n. 9), vi, discusses a's value in Lapo's quartet, but does not mention it when (xii) he comes to Arch. He may not have seen a himself, perhaps relying on Lagomarsini's famous collations of Florentine Cicero manuscripts. For a list of Lagomarsini's ‘codices Ciceroniani’, see Ruysschaert, J., Bibliothecae Apostolicae Vaticanae codices manu scripti recensiti: Codices Vaticani Latini. Codices 11414–11709 (Vatican City, 1959), xiv–xxvGoogle Scholar. The Arch. collations in ms Vat. lat. 11628 are described at 431–2.

46 In either case, the correctors seem to have acted in a rather superficial way, maintaining numerous untenable, even nonsensical readings. Whether they descend from a itself or from a's source, in either case b and c appear to do so through a common intermediary, as they share the errors solum nobis ad at 14.17 and adiuvaretur at 16.2.

47 Avignon, Bibl. mun. 1215, ‘Ex bibliotheca ff. Praedicatorum conventus Avinionensis’, is a fifteenth-century parchment manuscript of 87 folia (+ f. 37bis), and contains the three Caesarianae, the ‘thirteen’ Philippicae, Pro Archia, Red. pop., Red. sen., Pridie, Pro Milone and the beginning of De lege Manilia. I wish to thank Anna Bellettini, who provided me with a full description of the manuscript from the I.R.H.T. archives in Paris. A few illuminated capitals from A can be found on http://www.enluminures.culture.fr (accessed 12 April 2012).

48 A somewhat enigmatic case is Oxford's Balliol College 146B. This manuscript was copied for William Gray, probably in 1443 in Cologne: see Mynors, R.A.B., Catalogue of the Manuscripts of Balliol College Oxford (Oxford, 1963), 125–7Google Scholar. Mynors notes that ‘Billanovich 14 n. 27 suggests that this is a descendant of the copy of the pro Archia sent by Petrarch (its discoverer) to Boccaccio in 1351’. The reference is to Billanovich, G., ‘Pietro Piccolo da Monteforte tra il Petrarca e il Boccaccio’, in Medioevo e Rinascimento. Studi in onore di Bruno Nardi (Florence, 1955), 1–76Google Scholar, at 10–14 (and n. 27). The variants quoted above, however, seem to leave no doubt that the Balliol manuscript belongs to the α branch, instead of being a direct copy from Petrarch's codex for Boccaccio. Although in general its Pro Archia readings seem to qualify as an illustration of sloppiness and ignorance rather than of Petrarchan genius – as its text is utterly corrupt – it must go back to a deliberate collation process and probably a significantly less corrupt source after all. Among its numerous deviations, it combines the above-mentioned variants unequivocally typical of α with others that originated in β1 (such as: 8.11 et fide ] ac fide β1Bal; 12.22 defessae ] om. β1Bal; 13.7 quoque ] om. β1Bal) and that are too numerous to be coincidental. Therefore, it seems safe to conclude that, the Boccaccian appearances notwithstanding, Balliol 146B does not stem directly from Petrarch's desk. And given its heavily contaminated nature, if not its merger of two otherwise sufficiently attested families, it has no utility for the reconstruction of π anyway.

49 At 17.13, where P, Q and T only repeat the name of Roscius in the glosses, A adds ‘mimi ut puto’. At 27.11, Q alone repeats the name of D. Brutus in a marginal note. There A has ‘Decimus Brutus qui navali proelio Massilienses pro C<aesar>e superavit’.

50 P has the same note, omitting natura; T has incorporated it into the text after mentis viribus. The notes present in P have been published by Billanovich (n. 11), 103–4.

51 It would have been the only one, as the one other innovation shared by β and γ, that is the addition of the first sive at 30.12, is so much called for by the context that it would be hazardous to base any stemmatic conclusions on such an obvious correction.

52 Vat. lat. 1742 was copied from O before the latter was corrected, while BNF lat. 17154, B.L. Harley 5428, and Balliol 248A (the latter contaminated by the β branch) reflect O's corrected stage. The fact that the large Ciceronean collection BNF lat. 17154 belongs to the very restricted α branch might hold some promise for deciding the much debated origin of this important collection: see Hunt, T.J., A Textual History of Cicero's Academici Libri (Leiden, 1998), 144–6Google Scholar.

53 This α family largely corresponds to the ‘familia sexta’ in Elżbieta Olechowska's edition of Pro Cn. Plancio, Pro C. Rabirio Postumo (Leipzig, 1981), viii. While not without merit, Olechowska's grouping has serious shortcomings: her forgoing of any attempt at an eliminatio codicum descriptorum in combination with a rather vague dating, often limited to ‘saec. xv’, prevents us from seeing the wood for the trees. Out of the ten manuscripts listed as members of this sixth family, B5, O3 and P3 are in my opinion (and at least in Arch.) descendants of a twin of V4 (my O), from which descends β directly, while A2 and D5 are late (after 1450), corrupt and contaminated. As R3 and ζ in Arch. belong to the γ branch, the only two remaining relevant members of this family are my O and N. Likewise, other groups can probably be pruned significantly, freeing the stemma from quite a few of the dotted lines that may justly reflect contamination trails between families, but preclude a clearer view of the relationship between the more primitive witnesses belonging to each of them.

54 Another argument for the presence of alternative readings in β may be that M's probable twin, Plut. 48.10, has iussit at 25.30. The precise relation between Plut. 48.11 and Plut. 48.10, however, is somewhat problematical. Textual evidence surely excludes 48.10 as the source of 48.11, yet whether it descends from it is harder to prove or rebut. As Plut. 48.10 has iussit at 25.30, either β indeed had at least this double reading (but then, as iussit is absent from 48.11, the possibility must be excluded that 48.10 is a direct descendant from 48.11), or 48.10 is somehow contaminated, and then it is very hard to prove whether it is a contaminated descendant or twin of 48.11.

55 A detailed description of the manuscript is provided in the library's catalogue, Casarsa, L., D'Angelo, M., Scalon, C., La libreria di Guarnerio d'Artegna (Udine, 1991), 266–8Google Scholar (description by Mario D'Angelo). The manuscript opens with the pseudo-Ciceronian Fifth Catilinarian, for which it is the oldest preserved witness.

56 In addition to the presence in both manuscripts of Lapo's quartet in the classical order, D from textual units 17 to 23 (Pridie, Vatin., Red. pop., Har. resp., Catil., Ps.-Sall. Inv., Ad Alex.) has the same order as R. Moreover, in Cluent., both break off at 41.17: see Rizzo, S., La tradizione manoscritta della «Pro Cluentio» di Cicerone (Genoa, 1979), 123–4Google Scholar and 149–50. Both also contain pseudo-Demosthenes' Ad regem Alexandrum. This text is followed only in D by pseudo-Plutarch's Institutio Traiani. For the relation of this last text to Petrarch, see Zucchelli, B., ‘Petrarca, Plutarco e l'Institutio Traiani’, in Gallo, I. (ed.), L'eredità culturale di Plutarco dall'antichità al Rinascimento (Naples, 1998), 203–27Google Scholar.

57 Given the relatively late date of the genesis of D (compared to Petrarch's lifetime), I prefer to surmise the existence of an intermediate δ.

58 At any rate, the common origin of R and D that has been suggested (‘Le numerose analogie, segnalate da S. Rizzo [n. 55] con il Palat. lat. 1481, fanno pensare a una comune origine dei due manoscritti,’ D'Angelo [n. 55], 268) can only be contextual, and as far as Arch. is concerned not textual, as D's readings in our text show no affinity whatsoever with γ.

59 Ornato (n. 19), 565.

60 Ornato (n. 19), 565–6; Rizzo (n. 56), 33; Rouse and Reeve (n. 13), 87.

61 Unless one were even to hazard the hypothesis that Petrarch himself released the β copy of Arch. during his stay in Milan (1353–61), when he dedicated himself mostly to the study of the classics, especially Cicero: see Berté (n. 12).

62 It goes without saying that in a very complex and layered environment like the one characterizing the transmission of Ciceronian speeches in the late fourteenth and fifteenth century, one cannot jump to conclusions, from one stemma to another, that is. My reconstruction of the transmission may, for example, seem untenably at odds with the one presented in the 1992 Teubner edition of Quinct. – the only one, to my knowledge, containing an attempt to place all witnesses into a comprehensive stemma – as Reeve derives from my Q a mass of manuscripts that in Arch. display different dependencies. This contradiction is only apparent. When Quinct. came into circulation relatively late (at the end of the fourteenth century, see Rouse and Reeve [n. 13], 87), Arch. was already around in different families, where new apographs then added Quinct. Likewise, it comes as no surprise that the numerous manuscripts that Reeve traces back to his L (Vat. lat. 2903) in Arch. are spread over three families (α, β and γ), as Arch. is absent from L and L's descendants that do contain Arch. may have added it independently and therefore from different sources.

63 Novati, F. (ed.), Epistolario di Coluccio Salutati (Rome, 1891), 331–2Google Scholar: Sed ne, dum omnia quero, pluribus caream, hos vel in papyro transcribi peto: De lege frumentaria, Ad Hortensium, Pro Plancio, Pro P. Silla, De laudibus Magni Pompeii et Milonianam, quam ultimam habeo adeo corruptam et inexpletam, quod dici potest me illam penitus non habere. The order Planc., Sull., Manil., Mil. does not correspond to any of our manuscripts, but reflects the order of Petrarch's letter 12.8 to Lapo (quoted n. 7), with the only exception that there Mil. precedes the other three. It seems to make sense that Salutati knew Petrarch's letter and therefore tried to obtain the text from Petrarch's heirs, yet changed the order because of the specification he wanted to add about his own copy of the Mil. In 1370 he had already sent a copy of his Arch. to Tancredo Vergiolesi: Orationem Ciceronicam pro A. Licinio Archia egerrime manu mea exemplatam mitto (ibid. at 134).

64 See Rizzo (n. 12) for an overview of evolving views about this matter.

65 The bifurcation has been affirmed once more by, for example, Dominic Berry in his recent Pro Sulla edition (Cambridge, 1996), 72.

66 The importance of Q, as the most complete witness to Petrarch's emendations and other marginal annotations, should not overshadow its textual flaws. In fact, P is a more faithful witness to γ than Q, and their common ancestor γ3 ultimately represents only a limited part of γ.

67 Of course, early cases of Einzelüberlieferung and manuscripts containing one or more members of the quartet, and wanting the Pro Archia, deserve more attention than has hitherto been paid to them. In due time, I intend to test my stemmatic conclusions about the Pro Archia on the transmission of the Pro Sulla, the quartet member where the Itali branch is most important for the constitutio textus.