M. Tullius Cicero, Fasc. 43: De finibus bonorum et malorum. Bibliotheca Teubneriana

As recently as 1983 R.H. Rouse wrote “there is no adequate edition” (sc. of De finibus).¹ In the meantime the late L.D. Reynolds published his OCT.² When I reviewed it (this journal 2000.07.21), I would have been surprised to be told that another major series of classical texts would sponsor yet another edition within a few years; after all, 70 years separated Reynolds’ text from Martha’s Budé edition. Nor, though he once wrote an article on the manuscript tradition of Fin. (see below), would one necessarily have predicted C. Moreschini (hereafter M.) as the new editor, since he is known for work on (mostly Greek) patristics and late Latin, rather than Cicero. In spite of Reynolds, M. has brought to completion a project begun more than twenty years ago, spurred on by Scevola Mariotti, to whose shade this volume is dedicated (p.XV).

The volume consists of a preface (still in Latin in the Teubners) expounding the editor’s view of the relations of witnesses, a conspectus codicum, a conspectus editionum, then the Latin text with critical apparatus and between the two an apparatus supplying references for passages cited in the text and the indirect tradition. The lines of each Book are numbered continuously; in the following I cite the text by these numbers. An index of sources compiled by G. Duursma and an index of proper names conclude the volume. One misses a list of secondary works cited such as is provided in most Teubner editions and OCTs these days; instead M. provides references within the critical apparatus. But there are lapses; thus one must have recourse to Reynolds to discover publication details for Faber, cited by M. in the apparatus to 3.344, or H.F. Nissen, cited on 5.253-6.

To begin with the Praefatio: this is, as M. candidly admits (pp.V and χιιἰ, mostly a Latin summary of a paper he wrote on the subject in 1987.³ Hence the fairly detailed discussion of the manuscripts used by such older editors as Madvig, Baiter, Schiche, and Martha. But the elephant in the room is, of course, Reynolds; one would have expected an editor publishing this text within a few years to offer a detailed discussion of this predecessor in order to make the case that a new edition is needed. In fact, however, Reynolds’ work is discussed in two paragraphs (pp.XIII-XIV) describing some of the differences but merely asserting rather than arguing for M.’s approach. Nor does M. do Reynolds full justice when he describes his contribution as lying in an accurate list of earlier editions, in the selection of conjectures and in drawing up a clear stemma codicum (to which M. opposes his own stemma).

The analysis of the tradition by M. and Reynolds differs in several respects. The main witnesses to the text, namely A (Vat. Pal. lat. 1513, 11th cen.), B (Vat. Pal. lat. 1525 of 1467), and E (Erlangen 618, written in 1466), have been known and used by editors since the nineteenth century. These three are grouped together by M. as one branch of the bipartite tradition, whereby A has a direct line to the hyparchetype, alpha, whereas B and E derive from a posited intermediary (beta). Reynolds, however, sees B and E as forming a branch of the tradition (which he designates phi) separate from the other witnesses and cites (p.x of his edition) a number of errors in support of that view; one would have liked to see M. explain why he rejects this classification.

Another difference between M. and Reynolds is in the treatment of P (Paris. lat. 6331, 12th cen.): Reynolds groups it with another 12th-century codex, R (Leidensis Gronovianus 21), which has been studied in detail by Magnaldi.⁴ M., however, separates P from R and combines P instead with a group of 13th century MSS (LSY); he associates R, rather, with another 12th-century Paris MS, Pa (Paris. lat. 18104), which contains excerpts from Fin. ⁵ M.’s apparatus shows Pa agreeing in error with R alone at 1.236, 244, 610, 620, 660, 672, 677 etc. But this last passage argues that the relations are a bit different than M.’s stemma indicates: there viam is omitted by RPa and inserted (in the wrong place, following nullam in the preceding line) in PLSY. M. could explain the position by arguing that the word was omitted in delta, the hyparchetype above both gamma (from which derive RPa) and delta1, the source of gamma1, which in turn spawned PLSY (see M. p.XIV). But then he must assume that the word was also omitted from gamma2 (the source of MNV according to his stemma). But, in fact, the word is attested in the proper place in MNV. This and other indications shake my confidence in the delta hyparchetype as reconstructed by M. I suspect that the stemma should look more like that of Reynolds (his p.viii), with two branches alpha and phi (= BE) and under alpha the MS A as one branch by itself; one would modify his scheme by adding beneath his delta a common source (let us call it zeta) of R and Pa and then another stage (eta) beneath delta as the common source of PLSY. Then we merely have to assume that delta omitted viam at 1.677 and that the word was reinserted (in the wrong place) in eta.⁶

A final bone of contention is the classification of the Madrid MS (M = Matritensis 9116, 14th century): Reynolds combined it with two fourteenth-century Italian MSS (Mutinensis lat. 213 and Florentinus Cart. Strozz. 3.46; see his arguments, p.xi); M. admits the link but regularly cites it instead in combination with two fifteenth-century ones (N = Neapolitanus IV G 43 and V = Vaticanus lat. 1759). M. fails, however, to show separative errors of M against N and V and thus has not ruled out that N and V are codices descripti (he could, however, have adduced the material omitted by M at 2.1091-92 but transmitted by NV). In any case, the sorting out of this family will require more work on the part of a subsequent editor (though the text will probably not benefit much).

M. offers a generally conservative — often too conservative — text. It seems perverse to resist Lambinus’ habeant at 1.110, invenit of BE (1.232), Reid’s posited lacuna prior to itaque (1.189), C.F.W. Müller’s voluptatum pleniorum (1.577-8), Madvig’s ab eodem illo (referring to Epicurus; one expects a person in this construction) for transmitted ab eadem illa [1.697-8]), Orelli’s insertion of si for clarity at 2.278, Manutius’ praecipui for praecipue at 2.1264 (scribes were evidently deceived by its separation from the limited noun nihil), Ernesti’s clarifying ad portam at 2.1320, or Davies’ deletion of aerumnas at 2.1345; and Reynolds was surely right to prefer the more weakly attested indicari to iudicari, given the correspondence with aperiri in the preceding clause and the likelihood of an anticipatory error from the following iudicari (1.322; for the error cf. 2.1260 and 5.615 and 698). M. likewise follows the transmitted text at 2.142: vos de plagis omnibus colligitis bonos … viros sed certe non pereruditos. But have the Epicureans gathered their followers from “all regions” ( de plagis omnibus) or “all the country districts” ( de pagis omnibus)? In view of the antithesis, the latter, though attested only in a single witness, seems preferable (and is adopted by Reynolds); for urbani rustici as a global designation characteristic of Cicero’s mindset cf. 2.890. At 2.347 M. prints et quidem in the text, but equidem (SBE according to Reynolds’ app.; M. lists no variant), adopted by Reynolds, seems much more likely (the emphasis on illud is provided by the following ipsum; contrast 2.938). Surely iustus is not wanted in 2.679; it is likely to be an intrusion based on the following iniuste; Reynolds conjectured iniustus, unnecessarily: iniquus alone (βἐ, reinforced by the following quam multa vero iniuste fieri possunt …, will make the point. We surely need Baiter’s virtutibus, an easy saltation error following omnibus, at 3.88. Lambinus’ tam was easily skipped after tamen and aids the sense at 4.5; and Madvig’s ars seems a clear improvement over the vague res at 4.117. At 4.194 Madvig’s cum eum restores a plausibly Ciceronian structure (M.’s text involves asyndeton without a strong contrast before in utroque at 4.195) and is no real change (the tradition is divided between eum and cum). Manutius was surely right to delete nomina rerum commutantem as an anticipatory gloss (for a similar error, accepted as such by M., cf. 4.271-72). At 5.991 M. follows Reynolds’ judicious use of cruces; he might have done so also at 2.670 and 5.514-15; in the latter passage the unusual correlation et … atque should have put M. on guard.

Reynolds had a fine ear for rhythm: thus at 1.473 he read adhibenda est with BE to avoid the otherwise transmitted heroic clausula est adhibenda, which M. prints. The superior clausula also argues for etiam imitari velis of PLSY (double cretic with initial resolution) rather than the unrhythmical imitari etiam velis adopted by both M. and Reynolds; and M. should not have resisted Madvig’s insertion of est at 5.312, instantiating a double cretic; it can easily have dropped before the following et.

On the other hand, M.’s conservatism sometimes reaps dividends. He is probably right to reject the lacuna posited by Marsus at 1.224 on the assumption that Cicero meant to criticize Epicurean logic at greater length before turning to ethics; in fact the focus is to be on ethics with only a brief nod to the other “parts of philosophy” (cf. 1.294). At 2.469 Reynolds had adopted T. Bentley’s erit for the transmitted est to match the following futures; but he thereby destroyed a sentence-final cretic; est is therefore plausibly returned to the text by M. Again at 2.306 M. is probably right to see an energetic personal assertion ( hos ego asotos … dixerim) rather than a logical conclusion ( hos ergo …), pace Reynolds. Reynolds oddly preferred audaciter (BE) at 2.362 for the generally transmitted audacter, in spite of Quintilian’s having branded the former as pedantic ( Inst. 1.6.17); M. rightly sticks with the transmitted text. He likewise does well to retain transmitted referre, rather than adopt referri with Lambinus and Reynolds (2.1239), and the transmitted et at 3.740, changed by Madvig unnecessarily to sed. Careful rhetorical balance is, of course, a hallmark of Cicero’s style; hence M. does well to print et tuebere et munies at 2.977 (Madvig and Reynolds had omitted the first et); similarly at 2.1073 he rightly adopts et saepe plane diu to match the preceding nec … semel nec ad breve tempus. There is no difficulty about et … -que in Cicero;⁷ hence M. is right to repeal Reynolds’s deletion of et at 4.222.

In his philosophical works Cicero sometimes cultivates, for clarity, a fullness of expression bordering on the pedantic. For this reason I am not convinced by Madvig’s deletion of animo at 1.600 (the word is retained by M.). By the same token, however, M.’s deletion of voluptatum at 2.121 seems doubtful; I would instead follow Davies’ change of quae to qui in the previous line.

In general, the places where M. errs in changing the transmitted text are few: et indignantis adds liveliness to exclamations (cf. OLD s.v. et 15), but is often mistaken, as it was by both M. and Reynolds, who delete it (with Manutius) at 2.1180. At 4.14 M. follows Giusta in inserting dicere after soleo and prior to temere; soleo can, however, be used without an infinitive if the context suggests it (OLD s.v. 2a); here it would have to be supplied from the preceding oratio; but if the text is to be changed, one might prefer the order temere dicere as the likelier source of corruption (by saltation). esse must be supplied in 5.897-98 but not in the emphatic position following videre (so M., following Baiter); we should rather, with Madvig, insert it after honestatis, where it has the further advantage of effecting a ditrochaic clausula.

One finds few conjectures proposed by M. either in the text or the apparatus, but one of these is palmaris and will surely be adopted by succeeding editors, namely his nova for non at 3.162. He likewise did well to bracket dicere at 4.880 and is probably right to suppose that something is missing in the vicinity of 5.812 (though I would recommend inserting contemplans after est in 813, rather than legens after innumerabilia in 812). Future editors will want to consider his proposed insertion of a ratione after discrepant at 4.856, though they, like M. himself, will probably not regard it as absolutely necessary. At 2.867 M. conjectures vide for transmitted vides; this is not impossible, but in such ironic postpositive usage videlicet (conjectured here by Manutius, adopted by Reynolds) is more common and probably to be preferred.

In Fin. the indirect transmission does not contribute much; but at 5.705 Nonius’ hi might have merited at least a “fortasse recte” in the apparatus.

Other conjectures I would recommend for adoption: 2.524: Schiche’s insertion of contra eum for clarity, an easy omission after Chrysippum; 2.752: Wesenberg’s insertion of ut with addition of ita ( vino … ita ut ne noceret) to clarify that this is a consecutive clause with limiting force; 2.914 perhaps delete utilitatis causa expetenda as a gloss; 2.925 precarere (ed. Ven.); 2.1148-49 Baiter’s summa eius philosophiae (“the sum-total of his philosophy”) is surely what Epicurus’ will is said to contradict. In view of the following stylistic strictures at 2.1151 Epicurus can hardly be said to be writing breviter apteque; nor is aperteque, BE, adopted by Reynolds, any improvement in view of the following id eius modi est ut non satis plane dicat …; the solution will be Schiche’s arteque. At 2.1218 Schiche’s fluit is an improvement, an easy omission after preceding -re. I also recommend the addition of non prior to fuit at 3.822, the point being that Sulla was not master of those vices. Cobet’s tueri both aids the sense and obviates a heroic clausula at 4.73. At the very least we need to supply qui (= “how”) prior to posset dicere (4.268). The emphatic esse is surely needed at 4.379 (the better rhythm is effected by placing est with Gzius before rather than after profecto). Madvig was right to follow the recentiores in reading naturarum at 4.384 since here Cicero argues from the various naturae (= “creatures”; cf. my note on N.D. 1.26). At 4.458 one might consider hoc as the likelier source of the transmitted nos, rather than Lambinus’ ea. Scribes are prone to introduce the common verb habere in place of a less common one, hence the corruption (marked with a cross by M.) at 5.991; I suspect that similarly habeant is intrusive at 4.659 and that the true reading is moveant. haec can easily have fallen out before hactenus at 4.898; its insertion would restore the common formula. At 5.160 Strato is first called “great” ( magnus) in physics, but then the critique begins ( tamen, correlated with a preceding etsi); the first point should surely be that most of his points were “not new” ( non nova): the non was evidently omitted by haplography. Schiche was right to delete ista animi tranquillitas at 5.270 since ea ipsa provides sufficient reference back; the words will have been inserted by a scribe keen to explain the preceding εὐθυμία. At 5.630 one expects tam multa as the prompt for the following the consecutive clause, and the tam can easily have dropped out after quia. Surely the learned Cn. Aufidius, when he became blind, was moved by the litterarum lucis … desiderio (5.689-90). Ciceronian balance is enhanced by Davies’ Fragellarum for Fragellanum at 5.790. In the following line Cicero, as elsewhere, will have taken the opportunity to connect Athenian actors with the scene of the present discussion (cf., e.g., 5.1089 in hac urbe); Madvig’s huius is therefore to be preferred to R. Klotz’s suae — the former is also the easier error following quis. Another insertion of Madvig’s, illud dicere at 5.1000, is clarifying and can easily have dropped out following dicere (cf. also 5.986); and Baiter’s sint for sunt is needed in the same line for the apodosis within indirect discourse. Nor should M. have resisted Madvig’s elegant deletion of utentes in 5.869 and insertion of magis in 5.1048. We should probably also insert auditor before Polemonis at 5.1169.

The lively altercatio at 4.78 ff. has caused difficulties; Gigon thought 4.76-82 misplaced.⁸ Reynolds assigned the entire passage to Cicero. M. rightly divides it between Cicero and Cato, but then he really must follow Reynolds and the recentiores in writing agat in 79 (the subject is presumably Zeno, also the antecedent of ille at 4.82).

A special problem is the allusion to Panaetius at 4.275-81. He is reported to have written to Q. Tubero de dolore patiendo but without denying that pain is an evil. Transmitted is cuius quidem, quoniam Stoicus fuit, sententia condemnata mihi videtur esse inanitas ista verborum. I suspect that we need to change inanitas to inanitate and also quoniam to quamquam : although he was a Stoic (and therefore might have been expected to use the school’s characteristically sharp mode of discourse), his view was condemned by the emptiness of his rhetoric (i.e., in addressing the problem from the wrong angle).

M. uses fairly standard Teubner-style punctuation, i.e., he intersperses commas more liberally than Anglo-American editors are wont to do. This occasionally irritates; cf., e.g., the comma after explicemus at 5.775, which unhelpfully breaks the flow of the sentence; and Lucarini’s punctuation cited in the apparatus to 5.1150 seems a more natural way of taking the Latin than M.’s.

The brevity of the critical apparatus is a principle of the OCTs, whereas the Teubner series is more tolerant of prolixity. The contrast between Reynolds’ apparatus, a veritable masterpiece of concision, and M.’s could hardly be more striking. The latter could have cultivated crispness by substituting his siglum for the family where all members are in agreement on a reading; and many orthographica and peculiar errors of individual witnesses should have been omitted or relegated to an appendix so as to confine the apparatus to stemmatically significant variants. As it is, one must read that the first hand of A wrote herroribus in 1.148 etc. M. does, however, sometimes supply useful detail missing from Reynolds, e.g., on 2.423. On the other hand, occasionally the apparatus is too brief: when a negative apparatus is used with a number of different variants listed, it can be tedious to try to piece together e silentio the source(s) of the lemma (e.g. 2.549, 679).

The apparatus of citations is generally well managed (M.’s expertise in patristics shows itself to advantage in the reference to Jerome at 2.921-22), but there are omissions, e.g., of the passages from Aristotle and Diodorus cited at 2.235 and 237, though the references were already supplied by Reynolds. It seems doubtful that the Lucretius passage (1.74) cited on 2.103.1175-76 belongs in Duursma’s index fontium: probably both Cicero and Lucretius were thinking of the same passage of Epicurus.

I have noticed the following typos and omissions: p. XII, l. 17: observaverat (not observarerat); 5.841 praestigiis; app. crit., 1.145 quid (not quod); app. crit. p. 38: change 146 to 147 and add 148 before verbum; 2.626: the lemmata should be in reverse order; app. crit. p.100 change “287” to “286” and add “287” prior to accidunt; app. crit. p.121, l. 2 add “809” prior to “descriptus”; app. crit. on 4.2 a superscript 2 is needed after ut tam; inverted commas are omitted at 4.14 and elsewhere; app.crit. p.172 read “367-8,” not “376-8”; p. 176 app. crit. add “484” before subus.

To draw up the balance sheet: Reynolds’ De finibus, though very good, was not definitive; there was room for a new edition challenging Reynolds’ judgment on some points. M. does this but turns the clock too far back. The same Festschrift that included M.’s paper on the text of Fin. also included a paper by O. Gigon with a number of suggestions on the text, some of which were surely worth at least citing in the critical apparatus,⁹ but M. does not do so any more than he deals with the arguments of Reynolds’ Praefatio or the reviews of Reynolds’ edition. One regrets the apparent reluctance to go beyond positions taken up in 1987. The result is that those concerned with textual problems in Fin. will want to consult M. alongside Reynolds, but the OCT remains the basic text of this work, and a definitive edition remains a desideratum.

[[For a response to this review by C. Moreschini, please see BMCR 2005.12.14.]]

Notes

1. R.H. Rouse in Texts and Transmission: A Survey of the Latin Classics, ed. L.D. Reynolds (Oxford, 1983) 113.

2. M.T. Ciceronis De finibus bonorum et malorum libri quinque, ed. L.D. Reynolds (Oxford, 1998).

3. C. Moreschini, “Studi sulla tradizione manoscritta del De finibus di Cicerone,” Filologia e forme letterarie. Studi offerti a F. della Corte, 2 (Urbino, 1987) 252-67.

4. G. Magnaldi, “Il codex Rottendorfianus Gronovianus del ‘De finibus bonorum et malorum’ di Cicerone,” AAST 120 (1986) 133-60.

5. Cf. M. (n. 3 above) 256-61 for the relevant errors; Pa was already combined with R by Rouse and Rouse, as M. notes (p.256).

6. N2 has corrected from an exemplar of the eta type (cf., e.g., 2.757 and 1346), a special case.

7. Cf. R. Kühner and C. Stegmann, Ausführliche Grammatik der lateinischen Sprache 2 (Hannover, 1912, 4th edn. ed. A. Thierfelder, 1962) 36.

8. O. Gigon, “Überlegungen zum Gehalt und zum Text von Ciceros De finibus,” Filologia e forme letterarie. Studi offerti a F. Della Corte, 2 (Urbino, 1987) 235-46 at 245.

9. See previous note; e.g., his deletion of vel Epicuri at 2.443-44 in view of 450 has much to recommend it.