Bryn Mawr Classical Review 97.3.31

J.B. Hall (ed.), P. Ovidi Nasonis Tristia. Stuttgart and Leipzig: Teubner, 1995. DM 98. ISBN 3-8154-1567-5.

Reviewed by Peter Knox, University of Colorado, Peter.

In some respects Ovid's two collections of elegies composed after his relegation, the Tristia and Epistulae ex Ponto, fared worse than their author. For like him they have suffered a kind of exile in their exclusion from the canon of texts usually read in the classroom, but they have also had to endure additional punishment in the form of disfigurement at the hands of editors who have taken far too seriously Ovid's own criticisms of his exile poetry. Housman had harsh words for Owen's OCT, which appeared in 1915,1 but there was no satisfactory response to Housman's implicit challenge to improve on Owen's edition until the appearance of Georg Luck's edition in 1967, followed by a commentary completed in 1977. Now J.B. Hall has produced what will surely be the standard point of reference for scholars in decades to come. Since 1988 Hall has published a series of notes on the text of the collection in preparation for the edition, beginning with his participation in a colloquium in honor of C.O. Brink in 1987.2 Hall's final thoughts on the text of the Tristia are inscribed in this edition, which differs from Luck's in a staggering 500 places (the number can only be approximate), not including variation in punctuation, orthography, and the like.

When Housman took Owen to task, it was for his faulty editorial judgement and the poor quality of his conjectures; but he gave him full marks for the apparatus of his editio maior which provided full collations of the principal manuscripts. Hall, in turn, has been justly critical of Owen's contributions on this score, labelling him "one of the very worst collators I have ever encountered" ("Problems" 20). He went on to assert, rightly, "that what the Tristia still needs above all else, what indeed it has never yet had, is a full and accurate apparatus criticus, which at last will enable the reader to make up his own mind about the interrelationships and, more important still, the intrinsic worth of the manuscripts." An apparatus alone will not make good the need identified by R.J. Tarrant for a serious investigation of the textual history of the Tristia,3 but it would mark a significant advance. And so it is with relief that we find that Hall's commitment on this score has not faded in his edition, where he sets out his policies as follows (p. xvi): "cum tam male apparatui critico a prioribus editoribus prospectum fuerit, illud mihi ante omnia proposui ut apparatum conficerem quam maxime fide dignum." How fully Hall has made good on his proposition must be left to be determined by the next scholar to conduct a thorough investigation of the manuscript tradition. I can offer no better than a random sample of two of his "codices primarii". For Berlin, Deutsche Staatsbibliothek, Diez B Sant. 1 (B) I have compared his reports for Trist. 4.4a.1-4.10.132 with a microfiche copy and noted the following misreports over 414 lines: at 4.4a.14 the reading of B is posse where Hall reports saepe; 4.4a.36 mea (mea est); 4.5.14 ratam (raram); 4.5.28 iusta (nulla). In his description of this manuscript on p. x of the introduction Hall incorrectly lists the Tristia as occupying ff. 199-210; in fact they are to be found in ff. 199-217. Comparison with a microfiche copy of Tours, Bibliotheque Municipale, 879 in 5.10.13-5.14.46 (228 lines) produced more encouraging results with no errors identified. Hall's apparatus may be employed with the caution customary in such matters.

In the case of the Tristia, as in Ovid's other works, construction of a stemma, even if it were possible, would avail the editor nothing. As Hall notes in his introduction, some manuscripts display enough affinities to be identified as two groups, TrMV2 and AGHL4PV. But truth enters and departs the tradition with such randomness that the critic must rely on judgement, not a stemmatic crutch (p. xv): "consequitur ut semper et ubique ueritatem nil nisi ratio et res ipsa proclamet." Earlier editors, including Luck, were less than faithful to this precept with the result that good readings remained buried in the apparatus because they happened to have a home in the wrong manuscript. In numerous cases Hall restores to the text plausible variants which have rarely seen the light of day since Heinsius and Burman, although the Loeb edition revised by G.P. Goold is a notable exception: 1.1.12 passis, 1.1.14 sentiet, 1.1.17 illic, 1.1.124 morae, 1.3.2 quod, 1.4.5 increscunt, 1.4.25 uos parcite, 1.7.33 primi, 1.8.38 pede est, 1.9b.5 euenturum, 1.10.49 petantur, 2.8 demi iussa, 2.291 Iunonis, 2.305 quo, 2.331 dubito, 2.371 nisi turpis, 2.373 illic, 2.383 caecae, 2.501 hos, 2.521 nostris, 3.1.18 scribebar, 3.2.3 Latoia, 3.2.23 quod, 3.4a.6 arce, 3.4b.13 imago est, 3.11.21 diserto, 3.11.24 quamlibet, 3.11.62 Neptunique minor, 3.14.11 extremas, 4.1.42 modis, 4.1.48 hebet, 4.1.102 focos, 4.3.12 labat, 4.3.23 lectusque locusque, 4.3.57 proba, 4.5.28 nulla, 4.6.23 saepe, 4.6.48 male nocent, 4.9.31 receptus, 5.1.25 e, 5.2b.26 tutum, 5.3.50 deponat, 5.8.9 a, 5.10.19 aues, 5.11.13 fracta, 5.12.18 admissi, 5.12.28 uacauit.

Since Hall's judgement in such matters so often carries conviction, it will perhaps not seem churlish to remark that some neglected variants deserve their neglect. For example, at 5.14.39-40 Ovid invokes Laodamia in an epistle to his wife:

ut uiuat fama coniunx Phylaceia, cuius
Iliacam primo uir pede pressit humum?
Heinsius found primo pede in a fifteenth-century manuscript (Hall's V7 = Vat. Pal. lat. 1668), which he preferred against celeri of the rest of the tradition. Heinsius preferred primo "ex fide historiae," but that is a more likely explanation for why it supplanted celeri as a gloss. Ovid has already identified the protagonist by the epithet Phylaceia; celeri makes the point also made by ps.-Ovid at [Her.] 13.99-102:
hoc quoque praemoneo: de naue nouissimus exi:
non est, quo properas, terra paterna tibi.
cum uenies, remoque moue ueloque carinam,
inque tuo celerem litore siste gradum.
Protesilaus was quick when it would have been better to have been slow. Latinity, too, is not on the side of the variant, since primus pes means "the ball of the foot," and where's the point in that?

A second instance is found at 1.1.90-1, where Hall prints:

dum petit infirmis nimium sublimia pinnis
Icarus, Icariis nomina fecit aquis
Ovid is not given to expressions of the type "Grant is buried in Grant's tomb": the name of the sea is signalled by Icarus and Luck correctly notes ad loc. that Ovid does not repeat himself in such cases, aptly comparing Fast. 4.284 Icarus uastae nomina fecit aquae. Heinsius was wrong to prefer Icariis to aequoreis, demonstrating his unique aptitude for Ovidian word-play, shared with readers of Ovid in the thirteenth century, to judge from the variant ycareas found in 5 manuscripts at 3.4.22. The following variants elevated to the text by Hall might also merit re-evaluation: 1.2.69 laesos, 1.2.70 di, 1.2.92 uident, 1.3.92 uia, 1.5a.35 lapsis, 1.5b.18 sors tulit Sarmaticosque, 1.10.12 munere, 2.60 adiuui uotis, 3.1.74 ille, 3.3.21 deficiam, 3.4a.22 Icarias, 3.4b.11 errat, 3.5.31 quis enim, 3.10.11 nox, 3.11.54 querulos, 4.1.101 malignis, 5.3.42 Cressa, 5.12.26 equos, 5.13.30 peragat. It would also be worth hearing Hall's arguments in favor of beginning new poems at 1.5.45, 4.4.55, and 5.7.25, none of which seems likely on first inspection. His divisions at 1.9.37 and 5.2.45, although they appear more plausible, also stand in need of explanation.

Life would be simpler (and duller) if every crux could be resolved by resorting to a better variant, but when the tradition fails us and judgement alone can not repair the damage, we look for help to the critic's intuitive sense of the author to supply by conjecture what has been lost. It is in this area that Hall's edition is most likely to encounter some resistance. He states his position succinctly in the introduction (xviii): "amplissimum in apparatu critico coniecturis locum feci. mendosissimus enim est Tristium textus, nec nisi diuinando se res promouebunt." In a great many case Hall has made advances by recognizing the merits of conjectures proposed by his predecessors. In particular, the name of Nicolas Heinsius appears 22 times more than in Luck's edition attached to a reading in the text. Consider the case of 1.7.33-34, where Ovid suggests an epigram to be prefixed to his Metamorphoses. This is how the text is printed in modern editions:

hos quoque sex uersus in prima fronte libelli,
si praeponendos esse putabis, habe.
Too many readers have been forced to wrestle with the notion that Ovid could refer to the fifteen-book Metamorphoses as a libellus. Hall returns us to sense and Heinsius by adopting his simple primi in 33. But there are other names besides Heinsius to be found in the list of emendations which Hall has either reinstated or recognized for the first time: 1.2.101 quamlibet in (Diggle), 1.5a.7 quem (Diggle), 1.9a.35 ista ... uel ab (Heinsius), 1.10.43 hanc (Heinsius), 1.11.15 Azanidos (Alton), 2.111 aruo (Bentley), 2.125 electu (Palmer), 2.191-2 post 198 (Owen), 2.434 abest omnis (Rottendorphius), 2.437 quid (Owen), 2.495 nempe nec inuideo (Diggle), 2.542 inreuocatus (Bentley), 3.3.67 uetet (Heinsius), 3.9.20 aut (Heinsius), 3.11.61 felix (Housman), 3.11.71 timeam (Heinsius), 3.12.31 tandem (Schenkl), 3.12.41-2 post 36 (Wilamowitz), 3.12.52 iamne (Heinsius), 3.12.52 solum (Withofius), 3.13.3 seros (Burman), 4.1.9 ferens (Micyllus), 4.1.21 nec iter nec (Heinsius), 4.1.34 curae (Heinsius), 4.2.40 bene (Heinsius), 4.3.25 num (Heinsius), 4.3.41 certe (Burman), 4.7.8 frustra (Shackleton Bailey), 4.8.27 uacuum (Heinsius), 4.9.3 iungetur (Watt), 4.10.30 at (Schenkl), 5.1.29 at (Heinsius), 5.3.43 ut (Heinsius), 5.4.29 illi (Gilbert, Damsté), 5.5.24 consummetque (Heinsius), 5.7a.21 eheu tenerorum (Heinsius), 5.7b.25 male (Heinsius), 5.7b.41 mentemque (Shackleton Bailey), 5.10.11 sidera (Burman), 5.12.27 taetram (Heinsius), 5.12.66 usque (Gilbert), 5.14.12 suum (Shackleton Bailey), 5.14.34 rimatur (Heinsius). It is a safe assumption that the majority of these conjectures have now found a secure home in the text. And Hall's generous reporting of conjectures in the apparatus may bring to notice some others worthy of printing in the text: e.g. 1.3.92 uidua (Heinsius), 2.277 -que ortum uitium (Madvig), 3.4a.24 nonne (Heinsius), 3.8.12 -ue ... -ue (Bentley), 5.7a.18 trita (Housman).

Hall himself has published an astonishing 159 conjectures on the text of the Tristia, of which 127 are printed in the text of this edition. In a number of cases, his proposals are likely to win assent from most readers: 1.1.102 nostro, 1.2.81 nunc, 1.3.83 dicta, 1.3.102 absens, 1.4.14 ceruici equi, 1.5b.19 lectamque, 1.8.47 quamquam, 1.10.7 flatus, 2.6 mihi (jointly credited to Heyworth), 2.244 suadent, 3.1.74 ferimus, 3.6.7 caros amicos, 3.11.69 at quamquam, 3.12.2 iura remisit, 5.1.28 Musa mea, 5.1.52 grauem, 5.6.35 quodcumque, 5.12.35 surgunt. Unfortunately, however, the caliber of Hall's conjectures is not uniformly high. In his preface (p. xviii) he justifies liberal reporting of conjectures in the apparatus on the grounds that "usus autem docet etiam ex prauis coniecturis bonam aliquando posse prouenire solutionem." This statement contains a certain element of truth, but truly bad conjectures really don't teach anything useful and they certainly don't deserve a permanent residence in the apparatus, let alone the text. Hall's conjectures have for the most part been published before, albeit with a minimum of argumentation, since, as he puts it ("Notes" 125), his "firm belief is that if a conjecture cannot speak for itself, no amount of special pleading on the part of its propounder will suffice to vindicate it." This is neat; for it shifts the onus of proving that conjecture is needed, and if needed that the one proposed has the best claim to be true, from the critic to the reader. My own belief is that that is not fair and, in fact, constitutes dereliction of duty on the part of the textual critic. Life is short, argument long, but assertion is no substitute. An example: at 2.555-6 Ovid refers Augustus to his Metamorphoses as an example of his more serious poetry:

dictaque sunt nobis, quamuis manus ultima coeptis
defuit, in facies corpora uersa nouas
In "Notes" (1988) 130 Hall remarks, "I can see absolutely no reason why Ovid should have begun line 555 with dicta and ended it with coeptis." Accordingly, he emends to coeptaque, introducing a familiar Ovidian word play. But it is wrongheaded to attempt to impose a monotonous uniformity on Ovid's style and in this particular case Hall has gone spectacularly wrong because he has forgotten the opening words of the Metamorphoses:
in noua fert animus mutatas dicere formas
corpora; di, coeptis (nam uos mutastis et illa)
adspirate meis.
In Tristia 2.555-6 Ovid refers to the Metamorphoses by recalling its opening lines, and no amount of special pleading can vindicate an editor who proceeds on the basis of whimsy rather than reasoned argument.

The following examples of implausible conjectures printed in the text by Hall may perhaps illustrate that this is not an isolated case. Text and apparatus are printed as in Hall.

1.4.1-2 The setting of Bootes presages a storm:

tinguitur oceano Erymanthidos Vrsae
occiduasque suo sidere turbat aquas

2 occiduas- Hall: aequoreas- codd.
Hall, "Additional Notes" 132: "If I remind the reader that the final word of line 3 is aequor and the penultimate word of line 5 aequora, it will at once be manifest why I take exception to aequoreas- in line 2." But the word aequor and its derivatives have a curious tendency to turn up when the subject is the sea: there are fourteen occurrences in Met. 11.410-749 (Ceyx and Alcyone), three within the space of 7 lines at 11.427-33. The repetition alone is no objection to aequoreas (cf. E.J. Kenney, CQ 9 [1959] 248), especially since Hall's conjecture gives dubious sense. It is not only the western seas (occiduas ... aquas) made stormy at the setting of Bootes, but the seas generally, and Ovid is sailing to the east.

2.121-4 Ovid's house has collapsed, but not so grievously that it cannot rise again.

corruit haec igitur Musis accepta sub uno
sed non exiguo crimine lapsa domus:
ast ea sic lapsa est, ut surgere, si modo laesi
ematuruerit Caesaris ira, queat.

123 ast ea Hall: atque ea Cpc, cett. mei: ac ea P: at quae C ac V: atque ita L2O1: atqui 'cod. Richelianus' (=C) ap. Heinsium: quamquam ea Francius
ast is an archaism used by Ovid six times in the Metamorphoses and once in the Fasti. It used to appear at Pont. 3.2.70, but Richmond rightly eliminated it in his 1990 edition. It occurs nowhere else in Ovid, nor, for that matter, in Catullus, Propertius or Tibullus. atque admits of an adversative sense (HS 481) and the vulgate is unobjectionable.


en ego, cum patria caream uobisque domoque,
demptaque sint, demi quae potuere mihi

46 dempta- Hall: rapta H codd. demi Hall: adimi codd.
Hall emends because "rapta strikes me as rather too strong a word in this context, where caream has preceded" ("Notes" 131). But look back four more lines and the sense is plain (3.7.41) nempe addit cuicumque libet fortuna rapitque. Later at 3.10.61 pars agitur uinctis post tergum capta lacertis, Hall emends to rapta, presumably because capta strikes him as not strong enough, even though it suits the sense admirably, while rapta does not.

3.8.17-20 Ovid prays for Augustus to grant him return:

si precor hoc -- neque enim possim maiora precari --
ne mea sint, timeo, uota modesta parum.
forsitan hoc olim, cum se satiauerit ira,
tum quoque sollicita mente precandus erit

17 precari Fb1derU1: rogare A, cett. mei 20 precandus Hall: rogandus codd.

In 17 Hall prefers the variant precari, a plausible reading, which leads him to emend to precandus in 20 because "that would be very much in Ovid's manner" ("Seven Notes" 83). But so is the variation in expression brought by rogandus: cf. 3.1.77-8 di, precor, atque adeo -- neque enim mihi turba roganda est -- | Caesar. Hall imposes the same uniformity on Ovid at 3.13.25.

4.2.35-6 In the description of the triumphal procession for Tiberius' anticipated victory over the Germans, Ovid imagines an onlooker pointing out prominent captives.

illo, qui sequitur, dicunt mactata ministro
saepe requirenti corpora capta deo

36 requirenti Hall: recusanti codd.
Hall, "More Notes" (1990) 92: "Why should the German divinity in question be described as 'often refusing' the human sacrifice offered by the priest who now walks in the triumphal procession? If human sacrifice is acceptable at all, it presumably is the more acceptable the more the grisly quantity is increased." But it is not acceptable to Romans, hence the editorial recusanti. 4.4b.27, adduced by Luck, is precisely parallel -- deae ... crudelia sacra perosae in a reference to Artemis of Taurus. The Romans took it for granted that foreign gods did not differ essentially from their own (interpretatio Romana), and an aspect of their revulsion for the German rites of human sacrifice was the sense that it violated divine will (cf. Tacitus, Germ. 9.1). saepe should be construed with the entire predicate, not recusanti alone.

5.5.7-8 Ovid celebrates his wife's birthday:

quaeque semel toto uestis mihi sumitur anno,
sumatur fatis discolor una meis;

una Hall: ipsa V2 ulb 2: alba V2 cett. mei
What is the rationale for substituting una for alba? "It is only on the occasion of his wife's birthday, says Ovid, that he puts on a toga whose color contrasts with the color of his fortunes, and after that, we do not need to have the color spelled out to us If his intention was to wear a white toga only once a year, he would need one of them." ("Seven Notes" (1991) 84). Even if una could bear this interpretation, is it in Ovid's manner, as Hall might say, to register the tally of his wardrobe? Having stated that he wore the white toga only once (semel) is there a point in una?

5.8.19-20 Ovid caps a series of exempla of reversals of fortune with his own case.

nos quoque floruimus, sed flos erat ille caducus,
flammaque de stipula clara breuisque fuit

20 clara Ritchie: parua D : nostra cett. mei: rara Hall
In proposing rara in "More Notes" (1990) 96, Hall asks, "Is flamma nostra breuisque fuit really credible? 'The flame was mine and short-lived' is an extraordinary expression." So it is, but it is not Ovid's: -que connects the adnominal phrase de stipula and breuis, a relationship that is obscured just as much by clara, which Hall apparently now prefers to his original suggestion.

The number of instances where Hall has printed an inadequately supported conjecture of his own might easily be multiplied, and it is unfortunate that he has done the disservice to his readers of leaving to them the chore of removing them. Not even Owen did that. As Housman remarked, "The disfigurement inflicted upon Ovid's text by Mr Owen's recension is not a matter of dispute, for it is tacitly acknowledged by its author, whose labours on the Tristia for the last quarter of a century have chiefly consisted in removing his own corruptions and reinstating the comparatively pure text of his predecessors."4 One hopes that Hall will apply himself to the same task where it is warranted, for in the process he will do himself credit as well as Ovid, perhaps earning himself a place among the ranks of Ovid's better editors. One may also hope that he will then be led to reconsider some less fortuitous conjectures by others that now stand in the text: 1.1.74 quae loca (Richmond), 1.2.86 uelocem (Diggle), 1.2.99 merus (Camps), 1.3.39 nesciat (Burman), 1.8.21 faciunt bene (Francius), 1.8.27 haec (Heyworth), 1.8.48 summa (Ritchie), 2.231-2 post 236 (Ritchie), 2.281 saepta (Damsté), 3.1.63 peperere (Ellis), 3.6.15 mala fama (Ritchie), 3.10.65-6 post 56 (Ritchie), 3.12.46 signa (Heyworth), 4.4.28 nolis (Richmond), 4.4.45 per quem (Francius), 4.5.15 ingrato (Diggle), 4.8.6 non (Watt), 5.1.71 emenda (Wheeler), 5.8.30 rogata (Owen), 5.9.35 uetares (Broukhusius), 5.13.2 litore (D. Heinsius), 5.14.21 mea (Heyworth), 5.14.22 tua tum (Heyworth).

The publisher has included with the volume a slip of corrigenda, to which add the following: the note on 1.2.85 is misplaced in the apparatus; 1.9.35 in miseris is a manuscript variant, not an emendation; at 2.53 terras is missing its sigla; and at 3.4.6 the asterisk attached to arce in the apparatus should be deleted. In general the volume is attractively produced, although any subsequent edition should include a thorough overhaul of the typesetting in the apparatus so that entries are clearly separated.


1. Cambridge Review 37 (1915) 60 = Classical Papers 903-4.

2. "Problems in Ovid's Tristia," in Studies in Latin Literature and its Tradition, PCPS Supplementary Volume no. 15 (1989) 20-38, "Notes on various passages in Ovid's Tristia," Euphrosyne 16 (1988) 125-38, "More notes on Ovid's Tristia," Euphrosyne 18 (1990) 85-98, "Ovid, Tristia 2.77-80 and 5.11.25-28," LCM 16 (1991) 37-38, "Seven Notes on Ovid's Tristia," LCM 16 (1991) 83-84, "Additional Notes on Ovid's Tristia," Euphrosyne 20 (1992) 131-48. Throughout this review I refer to these publications by short title.

3. "Ovid," in L.D. Reynolds, ed., Texts and Transmission (Oxford 1983) 282.

4. Housman (above, n. 1) 60 = 903.