J.N. Madvig’s magisterial commentary on De Finibus (1839, 3rd edn. 1876) is a treasure-trove of fine syntactic and stylistic observations to which Latinists find themselves returning like a dog to a favorite bone. The edition to which it is attached is one of the first to be based upon application of the stemmatic method1 and represented a remarkable achievement in a time when travel to libraries was difficult and photographs unavailable. It could not, however, in the nature of the case, be definitive, and, though some incremental progress has since been made in sorting out the tradition, R.H. Rouse could state flatly in 1983: “there is no adequate edition.”2 Now finally we have from the pen of the late L.D. Reynolds (hereafter R.) a critical edition to take its place on our shelves as a fully worthy companion beside Madvig’s commentary.
Readers of this journal will be familiar with R.’s Medieval Tradition of Seneca’s Letters (1965), as well as his editions of the Letters (2 vols., 1965) and the Dialogi (1971) and of Sallust’s monographs, select fragments and the appendix Sallustiana (1991), not to mention the indispensable general introduction to textual transmission coauthored with N.G. Wilson, Scribes and Scholars (3rd ed., 1991), and the invaluable guide to the transmission of Latin texts which he edited and to which he contributed, Texts and Transmission, 1983. O.U.P. could scarcely have entrusted the new edition of Fin. to better qualified hands.
The chief of R.’s improvements is to the stemma. In his third edition Madvig had distinguished three classes of mss.: A, B+E, and a third category of deteriores depending upon a codex ignotus. With the greater resources at his disposal R. is able to refine this picture. He reduces the separate branches to two hyparchetypes, alpha and phi, the latter reconstructed from B and E, whereas A is now one of two sub-hyparchetypes under alpha, the other, beta, needing to be reconstructed as follows:3 (1) the delta branch consisting of Paris. lat. 6331 (s. XII = P) (Madvig’s codex ignotus, now identified) and its congener, Leid. Gronov. 21, s. XII = R, misplaced by Madvig among his codices meliores; (2) the gamma branch, reconstructed from three codices unknown to Madvig. R.’s contribution lies in clarifying relations of and within the beta sub-hyparchetype; see the stemma, p. viii.4 But even the oldest extant ms., A, dates only from the eleventh century and is very corrupt. Thus R. readily accepts that correct readings in individual manuscripts are more likely to be due to conjecture than tradition (p. xiv). What counts are the readings of the two hyparchetypes, individual codices being cited for the occasional interesting variant.5
R.’s Fin. is the second volume in a projected complete edition for OCT of Cicero’s philosophical corpus. One point shared with its series predecessor, M. Winterbottom’s De Officiis, is clarity of layout, whereby the needed information is set before the reader without the clutter of the dubiously relevant. The (Latin) Praefatio6 is followed by a list of editions and of “viri docti”, i.e., secondary literature. In the text orthography is regularized after Merguet’s Lexikon, as was done by Winterbottom (see below). Beneath the text is an apparatus identifying the sources of philosophers cited or the relevant attestation in the indirect tradition, then the critical apparatus itself, sometimes positive, sometimes negative, a model of judicious concision.7 The volume concludes with indices nominum (helpfully keyed to the relevant RE-articles) and locorum. Readers who compare this edition with Madvig’s will be grateful to R. for using inverted commas to articulate the parts of the different speakers; R.’s shorter paragraphs and more sparing use of the comma also promote readability. Dashes and parentheses are judiciously deployed to clarify the organization of Cicero’s sometimes intricate sentences (cf. 3.21, 35, and 40).
To improve on the text of so formidable a Ciceronian as Madvig is no easy feat, but R’s text does not suffer from the comparison (the examples cited here could be multiplied): 1.30: S’s indicari is rightly preferred to iudicari for its better antithesis; 1.37: the transmitted dictum sit is kept (no need to follow the recentiores in substituting est for sit); 1.53: C.F.W. Mueller’s voluptatum pleniorem efficit is a definite improvement on voluptatem; 1.61: Lambinus’ morosi is preferable to the transmitted monstrosi, which seems too extreme for this series of attributes; again at 3.12 Madvig was content to follow Manutius’ uti for ut in; it remained for R. to see that ut in has evidently been repeated from the previous line and to strike the in; and in 5.87 R. was right to print his own non quaerimus in preference to Davies’ non quaeremus adopted by Madvig.
An edition of a fairly extensive text such as this will, almost inevitably, present some small oddities, and R.’s Fin., for all its virtues, is no exception. Thus though R. says (p. xvii) that he is following the norm of Winterbottom’s Off. in matters of spelling, and Winterbottom in turn (p. xii) says that he is following Merguet, this is not carried through consistently, R. offering, e.g., parabilis (1.45), diis (1.62), and conbiberit (3.9), whereas Merguet has parabiles, dis, and combiberit. Then there is the question of Graeca : Merguet spells these (in an appendix) in Greek letters; Winterbottom, on the other hand, has chosen to follow the transliteration of the medieval manuscripts; but R. falls between two stools, sometimes transliterating, sometimes using Greek letters, with no underlying policy apparent; see, e.g., 3.13 (catalepsis) vs. the Greek spellings of 20, 23, and 69.
R. also follows Winterbottom’s OCT of Off. in another respect: Winterbottom revived from C. Atzert’s second edition (1932) the use of double square brackets to mark author’s variants not integrated into context. These also appear in two passages of R.’s text. This is a bit problematic though, since Fin., as the proem to Div. 2 makes clear, unlike, apparently, Off., was published by Cicero during his lifetime, so that any displacements in the tradition are presumably prearchetypal corruption, not author’s variants. Of the two passages, I think that 3.32 is better handled by M.R. Wright’s transposition to follow non contingit in 3.25, since this matter continues the discussion of the unique position of sapientia that precedes that point (for this reason Powell’s transposition to follow in artibus esse videamus [3.50] seems less convincing). On the other hand, I agree with R. that the words Iam membrorum… atque barba (3.18) belong neither here nor elsewhere in Fin., but, as hinted above, I would have used single, rather than double square brackets.
Given what is said above about the value and relation of witnesses, a decision where attestation is divided before a pause will often hinge upon the rhythm. R. is usually, though not always, alert to this criterion. Thus sapientia adhibenda est (1.43) effects, with elision, a clausula of the debuerat omne type and should be used (so, rightly, R.), not est adhibenda (alpha, followed by Madvig). On the other hand, Madvig rightly prefers vel summum vel ultimum vel extremum bonorum (phi), on the basis of its superior clausula, to the version in which bonorum is inserted after summum (1.42). Again, phi’s pendere sermone offers Cicero’s favorite clausula (type omne debetur) and would have deserved preference at 2.50 over the otherwise attested sermone pendere.
It seems only right that, just as the authors of conjectures are indicated in the critical apparatus, so the first to suspect a lacuna should be also. Thus when at 1.22-23 R. notes “post putat non pauca exciderunt”, the unwary might suppose that R. himself was the first to notice this, but, in fact, the suspicion goes back at least as far as Lambinus.8 On other matters one can rarely take issue with R., but I venture to suggest some alternative possibilities: at 2.100 it is said that Epicurus’ last letter is at odds with summa eius philosophia (so phi, adopted by R.) or summa eius philosophi (alpha), but surely what we need is Baiter’s summa eius philosophiae, Cicero claiming that the letter contradicts “the sum-total of Epicurus’ philosophy”, not Epicurus’ “supreme philosophy”, an unparalleled characterization. R. suggests ‘in possessione’ gloriae or ‘in usu’ gloriae as a solution for the genitives of 3.51 which otherwise have no connection to the sentence;9 perhaps one might consider ‘in’ gloriae ‘possessione’ : the order would then be parallel to the preceding in doloris integritate, possessione will have fallen out through saltation (e to e), and the now nonsensical in will have been dropped. Again, I wonder whether in 3.75 we might not need (of the sapiens): appellabitur … rectius magister populi (is enim est dictator) quam Sulla, qui trium pestiferorum vitiorum, luxuriae avaritiae crudelitatis, magister ‘non’ fuit as parallel to the preceding: rectius … appellabitur rex quam Tarquinius, qui nec se nec suos regere potuit. At 4.40 R. prints nam si ‘omnino nos’ neglegemus, in Aristonea vitia et peccata incidemus … sin ea non neglegemus… and for the daggered matter suggests in the apparatus ‘ea’ omnino [nos], thus positing two corruptions; I wonder if we do not rather need quae si omnino nos, the eighth-century abbreviation for nam by means of the single letter N written with an initial stroke below the line intersected by a horizontal stroke being not dissimilar to some versions of q(uae).10 In the same chapter I think Cobet’s aberimus an improvement over aberrabimus, which R. prints, the position of Erillus hardly being a standard from which “we” can be said to stray. At 4.60 R. prints: Zeno autem, quod suam [quod] propriam speciem habeat cur appetendum sit… and notes in the apparatus: quod2 seclusi; he might have noted that Baiter already bracketed quod propriam. One wonders, however, whether we may not have an instance of a double reading in the archetype with quod suam and quod propriam as the variants; in that case, quod propriam might be correct (cf. (sect) 75: illa argumenta propria videamus cur omnia sint paria peccata) and quod suam in need of bracketing. In 4.65 the transmitted text is: hebes acies est cuipiam oculorum, corpore alius nescit (or nestit B). R. follows Madvig’s senescit, but this restores a condition that cannot be treated, as the following development of the analogy requires; better to adopt P2’s languescit (albeit this is no less a conjecture). In 5.40 for ‘sitque’ extremum might one contemplate sic fit quod quaerimus extremum (cf. Fin. 3.29; Orat. 14 and 100; Leg. 1.18)? At 5.44 the parallel at 5.79 recommends T. Bentley’s ederetur for videretur. At 5.80 se fallat for ‘habeat’ ?
I have noted the following misprints: p. 5, l. 9: segnitiae; p. 45, l. 13: dolore; p. 108: in the app. crit. change “8” to “9” and “12” to “13”; p. 138, l. 17: change the second ob to ab; app. crit. 5.1: 1 audissem should not be italicized; p. 200, l. 17: iniuria.
1. For Madvig’s own contribution to stemmatics cf. Sebastiano Timpanaro , La genesi del metodo del Lachmann, new ed. (Padua, 1981), 50 and esp. 57-58; summary p. 78: “il primo a servirsi dello stemma codicum per la ricostruzione dell’archetipo non fu il Lachmann, ma il Madvig…”
2. R.H. Rouse in Texts and Transmission: A Survey of the Latin Classics, ed. L.D. Reynolds (Oxford, 1983; corr. rp. 1986), 113.
3. In his review of Madvig’s third edition T. Schiche, Phil. Verein zu Berlin 5 (1879), 186-201, already posited a bipartite transmission but failed to follow through on this theory in his own edition (1915); cf. R., p. ix, n. 15.
4. In view of the errors adduced by R., pp. x-xi, one need not harbor the dark suspicion voiced by Paul Maas, Textual Criticism, tr. B. Flower (Oxford, 1958), 47-48, that the trifid stemma is avoided so as to allow the editor free choice among variants.
5. R. has also provided a more ample account of the history of the text: “The Transmission of the ‘De Finibus'”, Italia medioevale e umanistica 35 (1992), 1-30.
6. This can no longer be taken for granted, even in editions of Latin texts: see the new OCT of Ausonius.
7. It would, however, be clearer to use positive apparatus when more than one variant is cited (see on p. 188, l. 4 for negative app. in such a case).
8. Those whose eyrie is in the Bodleian may be able to trace it still further back. — In his app. on 4.2, however, R. does clearly state “lacunam statuit Mdv…”
9. The latter after Heine, who inserted in usu elsewhere in the sentence.
10. Cf. Adriano Cappelli, Lexicon Abbreviaturarum, 6th ed. (Milan, 1973), 229 and 301.