[Ed. note: I have reason to believe that this is the first “issue” of a scholarly journal actually “published” in the main hotel at the APA annual meetings, by an editor who has just returned from the Michigan party, after warning the author of the review that his (first) fifteen minutes of celebrity were about to begin. The news from the APA is that the Goodwin award for best book of the year went to Susan Treggiari’s Roman Marriage. — Jim O’Donnell, 12/29/93]
Teubner (Leipzig) advertized a replacement for Koestermann’s edition some years ago.[1] The editors, B. and Kenneth Wellesley, were to divide the task along the lines dictated by the MSS: B. would edit the first Medicean, Wellesley the second (Annals XI-XVI appeared in 1986, the Histories in 1989). The set is now complete, not before the unification of Teubner and the raising of prices.
In his introduction, B. takes issue with Goodyear’s assertion that the recent trend in the editing of Tacitus has been to move closer to M. On the contrary, B. argues, recent editors have not followed M closely enough. B. describes his own practice in his note to II.22.1: equidem codicem Med. I religiosissime sequor. This is a risky stance: the text has clearly suffered from the diligence of an ignorant “corrector,” as well as from considerable carelessness. An edition guided by such rigid conservatism ought to exhibit two features: careful reporting of the contents of the mss, and strict (and reasonable) faithfulness to those contents. B. falls short in both areas; the contents of M are not always accurately reported, and B. does not adhere rigorously to any editorial principle. In what follows I address four difficulties with B.’s edition: inexact reportage of M and its marginalia; orthographic problems; insufficient attention to characteristic errors in M; and some possible bad judgement, particularly where Tacitean usage elsewhere advises against the reading that B. adopts.
M bears the marks of many pens, some early and some late. B. takes care to record some, but not all, of the marginalia, but only very rarely indicates whether he feels the hand to be early or late (some examples below). Even more problematic is his apparent confidence in some of these readings. My concern here is largely twofold: (1) it should have been possible to assign a rough date to many more of the corrections than he does, and (2) it should also have been possible occasionally to note where other corrections by the same hand could be found.
One particularly interesting comment passes entirely unnoticed by B., and is worth the attention of his readers. In response to I.58.6 (educatus Ravennae puer quo mox ludibrio conflictatus sit, in tempore me morabo), someone has written in the margin, in tempore memorabo | locatio Taciti in prin|cipio xi. B. does not here transcribe the contents of the mg. (nor, in fairness, are those contents noted in the commentaries of Goodyear and Koestermann). The author of this comment may, of course, refer to XI.16.1, from which it can be inferred that the son of Arminius has died. Or does he refer to some other passage? On the basis of this note, we ought to leave open the possibility that an early reader of M had access to Book XI and that his copy contained more of Book XI than we possess today. Unfortunately, the hand is extremely hard to date from the facsimile[2].
I.49.1 istem M | iidem (in mg. sinistro) | ijsdem (in dextro) (all as reported by B.). Here there is a series of difficulties. (1) The mg. on the left is hard to read. The hand seems to have written out iidem (the ending is abbreviated — a line over the e), and then either squeezed in a sloppy “s” to form iisdem, or that mark could form part of the attempt to cross this correction out. (2) The correction on the left being judged either incorrect or too sloppy, it is crossed out. The text on the right is properly read iisdem; the double “i” is identical (or very, very similar) to those at II.37.4 (Q. Hortensii in mg.) and III.29.2 (fastigii in mg.). At neither of those places does B. describe the double “i” as “ij”. (3) Finally, the hand in all three instances is certainly humanist, and likely that of Beroaldus[3].
II.39.1 fraude autuir aptum M | fraude aut vir raptum (M in mg.). Here B.’s note in the app. crit. does not suffice. The error is a simple one, and easily corrected. The success of our corrector here should not encourage any confidence — he missed similar howlers at III.22.3 (pertor menta M), IV.24.3 (recepto leameo M), and VI.4.3 (ne curas written out twice). Furthermore, this corrector underlined the text to be corrected in the mg.; a similar hand used the same practice at IV.10.2 and IV.14.3.
To my mind a certain caprice also informs B.’s attitude towards some of the abbreviations in M. M commonly offers in the place of quoniam the abbreviation quo (with a line over the u or o). In his expansive and helpful introduction, Rostagno restores this abbreviation as quoniam; editors had once used quando, but M always writes out quando in full. In most places B. follows the modern vulgate and restores quoniam. Several times, however, he writes quando (e.g., at I.59.5, without any defence). It is, of course, possible that sometime in the tradition both quando and quoniam were abbreviated and that some confusion between them has occurred; but B. nowhere puts forward this explanation.
The most troubling such passage is II.26.3. B. restores quo (with macron over the o) as quando, and claims the support of the margin where, to be sure, quando is written out. Earlier in the same paragraph M. offers regem que (macron over the e) aroboduum = regem quem aroboduum. In the mg. someone has written q; Maroboduu (macron over second u) = que Maroboduum. That correction, while sensible enough, is by a hand other than the copyist’s — but by the same hand which writes quando in the mg. below. The mg. therefore cannot support B.’s restoration. (Cf. also VI.6.1, where the mg. restores quo (macron over o) as quando, but all modern editors emend to quo modo.)
See also the comments below on the orthography of vecors and on the correction of prosperae at II.5.2.
B.’s arbitrarily maintained faith in M finds its clearest expression in his decisions regarding the regularization of the spelling of certain words and names. At the very least, an editor ought to separate such difficulties into two categories: those for which unanimity in other MSS and in epigraphic evidence offers a solution, and those for which we must determine Tacitean usage. On this issue an article by Ronald Syme has exercised a great deal of influence (“Personal Names in Annales I-VI,” JRS 39 (1949) 6-18 = Ten Studies in Tacitus, pp. 58-78; cited by B. 15 times). In that article Syme suggests that it is possible to determine which errors are the responsibility of Tacitus and which belong to the copyist, frequently positing that Tacitus has changed spelling with a change in source. This, it seems to me, is a dangerous position. Where should the line be drawn? Although Tacitus would probably have recognized alternative spellings of the same name or word, that is in itself no argument that he would have used different spelling haphazardly. Furthermore, to endorse the “source-theory” is to make a strong statement about Tacitus’s relationship to his source material. Yet Tacitus everywhere demonstrates independence of mind in style, selection, and arrangement.
I.55.1 Arminium. In 36 appearances in Books I and II M has forms of Arminius 33 times. Three times it offers Armenius (I.55.3, II.88.1 and 2). Standard practice (with which I concur) corrects to Arminius — as does B. Arminius is mentioned several times leading up to II.46; he is then absent from the narrative for a period, during which Tacitus refers several times to the Armenii. When Arminius reappears at the end of the book (II.88), what more natural than that the copyist should be confused? But a conservative should have noted that Armenius was an accepted spelling (cf. MSS of Frontinus and Florus, and the alternatives offered in Dio). It is, therefore, at least possible that Tacitus has here been caught changing sources; and a follower of the source-theory should, in the absence of certainty, have retained the paradosis.
II.1.2 Phraates. In Book II M offers Prahate, Prahates, and Prahatis, systematically corrected to Phrahate, Phrahates, and Phrahatis. The corrector did not cross out the “h” in any appearance (as at II.40.2 where aderant has the r crossed to make it adeant). In five appearances in Book VI, M has forms of Phraates (and in Book XII, M2 has Praatis, though Wellesley prints Phrahatis from the Leidensis). All these spellings are supported by ancient testimonia[4]. B. prints Phraates in Book II, ignoring both the corrector and the MS. This spelling has become the modern vulgate, and as such I accept it for the sake of clarity; but if Tacitus were anywhere to alternate spellings because of a change in source, why not here, where the word is a strange and foreign one?
M displays, and B. preserves, variation in the spellings of Syria/Suria, Treveri/Treviri, and Trebellenus/Trebellienus.On Syriait is hard to decide. But Treveri and Trebellenus are quite clearly correct, as proved by the testimony of contemporary epigraphic materials. There is plenty of evidence that M had trouble with names (e.g. I.10.4 and III.18.1, Iulius for Iullus; IV.9.2, Attus Clausus corrected incorrectly to Appius Claudius; IV.73.3, Cethecio for Cethego… and B. chooses to correct M in all these examples). More troublesome is the case of Latinius Latiaris (IV.68.2 and 71.1). When Latiaris appears again at VI.4.1, he is named Lucanius Latiaris. Syme urged an error on the part of Tacitus, and editors since have printed Lucanius. But M (as B. rightly notes) has been corrected at the second occurrence, to Latinius.
Again, B. accepts the arguments of Syme and preserves the erroneous Vescularius Atticus at VI.10.2 (even though Tacitus knew the correct name, Vescularius Flaccus at II.28.1); but, strangely, B. corrects M (from Vibulenus to Vibullius) at VI.40.1 on the evidence of Dio. Similarly, B. accepts the deletion of Visurgin at I.70.5 as introducing a geographical fallacy, but allows Cotye fratre to stand at III.38.2, when it is obvious from II.64.2 that Tacitus knew better. B. may have had a method of distinguishing the mistakes of Tacitus from those of his copyists, but its logic is neither self-evident nor explained.
M preserves similar variation in the spelling of certain words, notably aggerebatur (I.19.1, see Goodyear ad loc.); promiscuous vs. promiscus; percrebuit vs. percrebruit; and vecordia vs. vaecordia. Goodyear has a long and useful note on promiscus at I.48.1, in favor of the -us spelling. The uncorrected reading of M has that form 19 out of 20 times, and the correction to -uus is hardly universal. B., for whatever reason, accepts the correction to -uus where it appears, but otherwise prints the -us form. For no apparent reason B. changes his practice and regularizes vaecordia to vecordia, even though vae- is the spelling offered by M in 4 of 8 appearances. Furthermore, in 2 of the remaining 4 ve- has been corrected to vae-. This correction (a descender from the e, extending to the left and about half-way to the next line, with an ascender almost forming a loop under the e) appears many times in M[5], and, although he universally accepts the correction elsewhere, B. nowhere acknowledges it as such in his app. crit.
Compare B.’s reading at I.65.4: en Varus et eodem[que] iterum fato vinctae legiones. Why delete que or et? To do so requires that we also hold suspect Hist. i.80.2, ii.2.1, iv.53.4, and iv.54.1. (At i.80.2 an early corrector deletes et.) Not helpful is the fact that reporting on these lines in recent editions has been somewhat haphazard. Actually, I concur that the redundancy needs correction, and B.’s note is full and helpful. But all those who agree must admit that this evidence suggests that at some point in the tradition (at least of M2) a copyist felt free to introduce a systematic “correction.” Could not the same have happened with the orthography of promiscus?
Not surprisingly, the scribe of M displays certain tendencies in his errors, and proposed emendations of M could frequently be defended or rejected on the basis that a similar error has been committed elsewhere in the text, one for which modern editors agree that emendation is necessary. Yet, in defending the paradosis, B. frequently disregards, or simply fails to mention, the characteristic nature of the suspected error.
I.7.1 adulatione M, B | adulationem Heinsius. The emendation restores both sense and balance, and obviates the need to understand a different verb with lacrimas gaudium. B. cites examples that support either the MS or the restoration, but in his examples the verb has two unmodified objects; here there must be parallelism between lacrimas gaudium and questus adulatione<m>. Furthermore, the first word on the next line begins with an “m”; such confusion resulting from the ending of a line can be seen at I I.5.2, where M reads prosperae/evenissent: the “e” on prosperae is marked with three dots, as to be deleted (not a single dot, as recorded in B.’s app. crit.). (Cf. the dittography at IV.73.2, where M has pelluntur|turmas = pellunt turmas.) At I.7.1 the scribe simply did not notice his error.
I.32.1 sexageni singulos M, B. In truth, the received text could probably carry the meaning which Goodyear (with Furneaux) prefers, and to achieve which they accept the emendation to sexagenis. But their objections to the meaning that sixty men beat each centurion are valid. Would the received text have been ambiguous to a Roman? Perhaps, and the ambiguity would serve no purpose. The change is an easy one, what with all the s‘s around. I favor emendation. Cf. I.75.4, where M has causa senatui, corrected by Sirker to causas senatui. M also makes errors in the other direction: e.g. IV.21.2, where an extra “s” joins secretissermonis = secreti sermonis.
B.’s practice regarding other textual difficulties is often equally problematic, and at times equally haphazard. Frequently he retains the paradosis where it is grammatically acceptable but either confusing or unnecessarily ambiguous, or at times when it offers an unnecessarily harsh style. In particular, his defence or denial of various emendations inspires unease, allowing deceptive parallels in other authors to take precedence over Tacitean practice.
I.19.2 incipientes M, B. We should almost certainly correct to incipientis, a very easy change which Koestermann defends persuasively (Hermes 66 (1931) 477-478). Some of the “parallels” cited by B. are not really parallels at all — no one really doubted whether incipiens could modify cura, and therefore examples of incipiens modifying seditio or amor are useless.
IV.65 auxilium <por> tavisset Doederlein, B | auxilium tulisset Lipsius. Doederlein (with B.) compares Sall. Cat. 6.5, but again there is no reason why Sallustian examples should prevail over Tacitean usage. auxilium is not used with porto in Tacitus, itself not a common verb in Tacitus. But see tulere auxilium at III.41.2. Cf. I.19.2, where B. would keep M’s desideria … ferendum. His defence, by comparison with Plautus, Lucretius and Vergil, is admirable, and it would be persuasive had he found a Tacitean precedent. But Tacitus elsewhere declines gerundives, including those of fero; and therefore I can see no reason for retaining such a rare archaism here[6].
I.41.1 pergere ad Treviros et externae fidei M, B. B. (with Heubner and Koestermann) retains the text. Goodyear’s note ad loc. is thorough. His conclusion can be stated briefly: pergere does not take a dative in Tacitus. To retain the text we must therefore postulate an ellipse of, for instance, homines. As it stands, however, the text is quite difficult to interpret. Much is solved by the suggestion of Walter, externae fidei <dedi>, an easily explained error. At the very least the sentence ought to have been obelized.
II.61.1 nullis inquirentium spatiis penetrabilis B. | nullis inquiren<ti res>tium spatiis penetrabilis Goodyear. Beyond accepting (as everyone does) Lipsius’ penetrabilis for M’s penetrabiles, B. retains the paradosis. But we should not be mislead into accepting it simply because we have read Herodotus and therefore read into the text what we think Tacitus intends. spatiis is too vague, and nullis inquirentium not particularly elegant. Goodyear’s suggestion, which B. labels ingeniose, is just that[7].
Equally problematic are instances in which B.’s punctuation suggests either that he has imposed order where none is to be found, or that he has misunderstood the emendation which he accepts. A thorough discussion of the problems would make this review too long; here I simply caution the reader in particular about II.8.2 and II.33.3. Note that the fundamental treatment of II.8.2 by Heubner appears in Gymnasium 63 (1956), not in Gnomon, as written in app. crit.
B. offers several suggestions of his own, and occasionally his notes are quite full. I mentioned above I.19.2. Also admirably full are notes on III.21.4, IV.44.1, and VI.4.2, although I find none of these ultimately persuasive. At III.3.3 his minor change is nice, and is defended by Tacitean practice. At IV.31.3 his emendation is not supported by the passage he cites, since the grammatical structures of the sentences are not analogous[8]. At VI.15.3 he writes a long and helpful note which, insofar as it defends the modern vulgate, is perhaps unnecessary. At I.41.1 (a notorious passage) M offers quod tam triste.Koestermann accepts the paradosis, but most editors have urged emendation. B. changes to quo tam triste <iter>, a very attractive solution which he claims as his own (scripsi). But <iter> had been proposed by Jacob, and Ronald Martin seems to have supplemented this with quoin a suggestion to Goodyear some years ago (see Goodyear’s note ad loc., p. 283 n. 1).
The quality of the production is relatively high. B. has chosen to use continuous line numbers throughout each book. The app. crit. is complicated by his decision to describe corrections in M, but nevertheless there are few mistakes. At I.57.1 (l. 786) the cross-reference should be to I.59.5, not 59.6; at II.2 no chapter number is printed in the margin; at III.35.3 the app. crit. refers wrongly to line 486 (read 466); at IV.21.2 lines 291-292 the reference should be to JRS 46 (not 26); at IV.31.3 line 415 the word “scripsi” in the app. crit. should be italicized; at VI.7.1, 9.2, and 10.2 the citations from Syme should not be to his Tacitus, but to his article cited above; at VI.40.1 the reference should be to JRS 39 (not 31). Paragraphing is quite conservative. What is more, B.’s volume cannot stand alone: where Wellesley provides an exceptionally full bibliography in both his volumes, B. merely refers the reader to recent bibliographical publications; and the very full index to the entire Annals is included only at the end of XI-XVI.
B.’s text has a niche: among currently available texts it offers the most complete app. crit. But B.’s view on the quality of M seems to me to render his text highly problematic for the unsuspecting reader, not least because few editorial principles have been consistently applied. Instead, such principles as emerge are used or jettisoned according to their usefulness in retaining particular readings. Even with all its correctors, a startling number of obvious errors remain in M; and in light of that fact it is hard for me to see why we should trust M where the problems are most complex.
[1] For my own readings of M I draw on the facsimile reproduction in Codices Graeci et Latini photographice depicti, vol. 7.1, with an introduction by H. Rostagno. Prof. Charles Witke of the University of Michigan provided aid in the reading and dating of some marginalia, and Prof. A.J. Woodman of the University of Durham read through a draft with exemplary patience.
[2] It has previously been suggested that one ancestor of M2 may have been a MS in minuscule with approximately as much text per folium as M1. This note supports the possibility that one copy of the Annals (of undetermined completeness) may have come to Montecassino from Germany.
[3] Cf. I.48.2 eas caenia M | eas Cecinna (M in mg.) Here B. does not fully reproduce the text in the mg.; he records the mg. as containing only Cecinna. The corrector found the beginning of the sentence unclear and wrote it all out, with a correction — idiotically, since the correct name (Caecina) appears in the previous sentence. Moreover, the hand is one of the latest in the mg. of M, very likely post-Renaissance; but B. does not mark it as late at all (as he marks V.7, for instance).
[4] Res Gestae 32 Phrates (same spelling in some MSS of Val. Max.); at II.91.1 the editio princeps of Velleius has Phrahates, but a contemporary copy from the same exemplar reads Prahates. Cf. Hor. Carm. II.2.17 Prahates, with Nisbet-Hubbard’s note ad loc.
[5] I.47.1 (Italia to Italiae), I.47.2 (quedam to quaedam), I.48.1 (presumant to praesumant), I.59.4 (que to quaeque to quae), II.4.3 (Syrie to Syriae), II.28.3 (male to malae), II.30.1 (vecordes to vaecordes), II.72.1 (que to quae), III.50.3 (vecordiae to vaecordiae), VI.4.1 (luende poene to luendae poenae) and VI. 4.3 (emulationem to aemulationem). This list is by no means complete. At II.28.3, M has fame male statim corrected to fame malae statim. B. prints famae malae statim without any note in the app. crit. The correction is extremely hard to date. It may have been made with a thinner pen, but I can say no more on the basis of the photograph.
[6] Cf. IV.46.1 incultu M, B | inculti Beroaldus. That the noun incultu should occur in Sallust is simply no argument for retaining it here, when the arguments for inculti are so strong. See Martin-Woodman ad loc. Their commentary on Book IV (Cambridge, 1989) seems not to have been consulted by B., and I therefore refrain from discussing their own emendations. I note here that I support B.’s claim concerning IV.8.4: I see no evidence in the facsimile that conformaret is the result of a correction from confirmaret. I refer the reader to particular fine suggestions by M-W at IV.21.3 and especially at 28.1.
[7] Cf. IV.26.2 repetitus ex vetusto more missusque e senatoribus B | repetitus ex vetusto more <honos>, missusque e senatoribus Doed. Doederlein’s supplement, now the vulgate, is necessary. repetitus cannot stand alone so close to missus, with which an entirely different subject must be understood. Cf. also VI.26.1, where B. accepts continuus principis (defended by Draeger, among others) as short for continuus sodalis or amicus. But I can find no acceptable example to support this usage. Cf. also V.4.2, where B. retains festis (most restore faustis, comparing I.35.3 and IV.9.1) and compares XII.69.1, where, ironically, Wellesley has restored faustis.
[8] At I.14.4 the se was the subject of an indirect discourse. Here se would serve no such purpose.