P. Ouidi Nasonis Metamorphoses. Oxford Classical Texts

Some are transformed just once
And live their whole lives after in that shape.
Others have a facility
For changing themselves as they please.
— Ted Hughes, Tales From Ovid

From the bright lights of the Broadway stage to the dimmer ambiance of the seminar room Ovid’s book of changes continues to engage, fascinate, and inspire audiences with tales from the volatile world of his hectic invention where bodies human and divine play out their epic scenarios of love, anger, and transformation. In recent decades Ovid’s carmen perpetuum has generated an immense amount of critical and creative activity in the form of commentary, interpretation, translation, and adaptation. And the source of all that activity is ultimately the material text of the Latin poem, the inescapable embodiment of Ovid’s poetic consciousness transmitted from his own day down to ours. All the wonderful qualities we associate with the poem, wit, sophistication, fluency, brilliance, narrative energy and ingenuity, originate in inked marks on the page; so it is important to recognize, if only for a moment or two before becoming immersed in the glorious make-believe of Ovid’s incomparable tales, that our construction of his poetic genius in the Met. is in large part the result of a centuries-old, collaborative editorial effort that has investigated and sorted out the manuscript tradition, and has established the poet’s meaning on the basis of that tradition, or restored it by conjecture where the tradition is deficient. We are the beneficiaries of that collaborative effort.

For nearly thirty years now W. S. Anderson’s critical edition of the poem, first published in 1977 and now in its ninth printing, has been a bestseller in the Teubner series and has provided the Latin text used in countless reshapings of the poem by critics, commentators, and writers. Despite shortcomings that resulted from too great a dependence on the authority of the primary witnesses, Anderson’s edition has held the field virtually unchallenged (though there are those who prefer G. P. Goold’s revision of F. J. Miller’s Loeb edition [1977]). It now has a rival.

R. J. Tarrant’s long awaited edition of the Met. has appeared and will undoubtedly establish itself as the critical text that takes us from the twentieth into the twenty-first century of textual and literary scholarship on the poem. Readers who are familiar with the pre-publication version of his text printed alongside D. E. Hill’s fine translation will discover that their high expectations are here met or exceeded. Of editions of the Met. that are available to readers this is the only one that combines a full and accurate report of the manuscript tradition, presented in a well organized apparatus, with a text that is the product of, in the editor’s phrase, “an enlightened eclecticism” which accepts readings found in the twelfth and thirteenth century manuscripts, and is more open to conjectures than the conservatism of the twentieth century editors. The tyranny of the primary manuscripts, which was once strong enough to maintain the unmetrical Laertiadaeque (12.625) or the repetition of micant (3.34), to give just two examples, is over. Now Ovid’s brilliance will sparkle all the brighter.

The praefatio (v-xxxviii) gives a clear overview of the tradition, descriptions of forty-five manuscripts, fragmentary or complete (not counting codices deperditi), and a discussion of editorial methods. Tarrant has added a new witness to the ranks of the antiquiores, G, Sangallensis 866, s.xii, which lacks 8.548-10.428; it had previously been overlooked because it was wrongly dated. G contributes some new readings, or supports readings previously attested either in a minority of the primary manuscripts or only in the recentiores. Some of the more interesting of these are: 1.231 domino; 2.871 primis; 7.30 segeti; 7.786 morsus; 10.637 quod; 14.508 subitarum; 14.648 iurares; 15.502 finxit uoluisse; 15.729 omnis. The introductory matter also contains a list of editions (xxxix), a bibliography of textual studies (xl-xliii), and a helpful comparative table of manuscript abbreviations (xlvii-xlviii) used in the present edition and in the editions of Anderson, Magnus, and Slater ( apparatus only).

In the praefatio the editor, who in previous publications has contributed much to our understanding of the poem’s transmission, concentrates on three points: first, the tradition is a contaminated one in which comparison of manuscripts during copying and readers’ annotations generated by the frequent reading of this popular work led to the promiscuous exchange of readings among copies, good for bad and bad for good, so that no “pure” strain of the textual tradition can be identified; second, since, as a result of pervasive contamination, no one manuscript or family of manuscripts deserves preferential treatment of its readings, it must be recognized that good readings can turn up in the recentiores of the twelfth and thirteenth centuries; as the editor himself well observed elsewhere, “… in the tradition of the Metamorphoses every lectio recentior is potentially a lectio vetustior for which an early witness does not happen to survive” ( Formative Stages of Classical Traditions (1995) 114); third, the most effective editorial method for dealing with this particular set of textual circumstances is eclecticism. Although this may not be exciting news to those who have been following the investigation of the transmission, there will nonetheless be considerable excitement when readers see the results of that method. Tarrant has cast his net over the wide and treacherous sea of codices antiquiores and recentiores, and scholarly conjecture; the harvest is rich and impressive.

Although eclecticism is the method employed for the evaluation of readings, the editor offers a stemma codicum (xxvii), with the caveat that it is not to used for the reconstruction of the readings of an archetype; contamination will not allow it. To avoid the suggestion of an archetype, the symbol Ω, which represents the consensus codicum, is omitted from its traditional spot at the head of the stemma. Ω will represent not a text found in a single codex, but rather a text common to multiple codices no longer extant; each of these in turn will have contained variant readings. Multiple copies with variants will account for the pervasiveness of contamination and the great number of variants found among their descendants. The value of the stemma lies in its usefulness as a general outline of the transmission and as an organizational device for reporting the manuscript evidence in the apparatus. The frequent occurrence of the editorial comment fort. rect. beside readings in the apparatus is a clear reminder that as an instrument for the evaluation of variants in the Met. the stemma is a map that leads to a dead end. It is also a reminder that the editorial problem in the Met. is the problem of eclecticism in general, the limitations of stylistic and aesthetic criteria, and enlightened judgment for selecting the authorial reading. A reading found in the recentiores that is judged to be Ovidian (e.g. nomen 4.417) may be no more than a clever sophistication of the reading found in the primary manuscripts ( numen).

The stemma reproduces, apparently through some oversight in production, the stemma in Tarrant’s article in Texts and Transmission ([1983] 282). As a result, not all the manuscript symbols used in the apparatus criticus and listed in the Sigla (xliv-xlv) correspond to the symbols used in the stemma because the editor has introduced new abbreviations for the fragmentary manuscripts, formerly designated by small case Greek letters, and for a few other manuscripts. The OCT stemma was not updated to include these changes. The most serious confusion results from the new symbol E; it now represents Additional 11967 (formerly β), a manuscript in the Δ family. On the OCT stemma E represents Vaticanus Pal. Lat. 1669 (now rechristened π a member of the Σ family. The confusion is easily avoided by penciling in the correct sigla with reference to the comparative table on xlvii.

On xxxiii-xxxv the editor takes up the questions of interpolation and Doppelfassungen in the text. Both questions are too large and intricate to be reviewed here. Tarrant has made a major contribution to the study of interpolation in Ovid, especially with his development of a typology for the classification of interpolations by kind. His publications in this field, which readers should consult for his views on specific lines, put the detection of spurious verses on a surer footing. By my count one hundred twenty-two lines are bracketed in the text; six are deleted, five of which belong to a single passage, 12.434-438, in which readers are treated to the picture of Tectaphos’s gooey brains oozing out of his nostrils, eyes and mouth (a picture worthy of Lucan); another thirty-four lines labor under the cloud of editorial suspicion. This total does not represent a high percentage of interpolated and doubtful verses in a long and popular text that attracted readers in all stages of the transmission, some of whom were no doubt inspired to collaborate with the author. I am sad to see 2.226 in brackets, aeriaeque Alpes et nubifer Appenninus, though in the article cited below (428) Tarrant makes a convincing case against its authenticity. The lines that are bracketed, deleted, and branded suspecti are conveniently assembled in an appendix to the editor’s paper in HSCP 100 (2000) 425-438. With regard to the problem of those groups of verses that may represent a “double recension” by the poet, the editor states his position clearly: only one form of the text is genuine, the other is an interpolation (xxxv). In practical terms this means that a single Ovidian version is fashioned out of these controversial passages by means of deletion and rearrangement. The treatment of this highly complex aspect of the transmission in the apparatus is exemplary.

Before leaving the praefatio I would like to call attention to two things. With regard to the collation of the witnesses, the editor writes, “Omnes [testes] ex imaginibus lucis ope factis primum contuli, sed quosdam codices ipsos adii ut locos lectu difficiliores propius inspicerem” (vii). That adds up to a lot of hours feeding the microfilm reader and eating the dust of parchments. With regard to the overall condition of the text of the poem, the praefatio may leave impressionable readers with the optimistic view that the text is in good shape. To counteract that innocent optimism I quote two others who have studied the transmission: first Slater ( Towards a Text ([1927] 38), “stabilem horum carminum textum et integrum non habemus”; and then Richmond ( Brill’s Companion to Ovid ([2002] 470), “The loss of M N after Met. 14.830(838), and of the first hand of U after 15.493, gravely injures the textual state of the end of the poem” (470).

Readers will want to know what makes the text of this edition different from Anderson’s. The short answer is that Tarrant’s eclecticism, which embraces both the primary manuscripts and the recentiores, and his willingness to accept conjectures not only in places where the tradition is manifestly defective produce a better text. (The long answer will be found by collating Tarrant’s text against Anderson’s; samples will follow.) In this respect, I think it fair to say that the text of the OCT has more in common with the editions of Merkel (1875 ²), Korn (1880), and Riese (1889 ²) than it does with the editions of Magnus (1914), Ehwald (1915), and Anderson (1977). While the latter editors took the text of the primary manuscripts as far as it can go, and in so doing revealed just how unsatisfactory the result is, Tarrant has followed the more critical and the more difficult approach of his nineteenth century predecessors mentioned above. To give the reader a good idea of the general character of the edition, I offer a sample comparison of variants and conjectures printed by Tarrant with those printed by Anderson. In my judgment, the text of the OCT is superior in the following instances and reveals the high quality of Tarrant’s text. In all that follows the text of the OCT is given first. There is no point in repeating the manuscript sigla here; a reading is either a transmitted variant, or a conjecture identified by the name of its author. I do, however, indicate readings found only in manuscripts of the twelfth and thirteenth centuries or later with an asterisk in order to show that the editor does indeed practice the eclecticism that he preaches. Those who want to know more about the manuscript sources of the readings should consult Tarrant’s apparatus.

1.53 pondus aquae leuius] pondere aquae leuior; 1.439 incognita] incognite (at last the gender problem of this snake is solved); 1.601 Argos L. Müller ] agros; 2.38 generis] genitor; 3.34 tres ( Bothe) *uibrant] tresque micant; 4.610 *Iouis ¹ ] deum; 4.647 *moenibus] montibus; 5.181 miracula] oracula; 5.460 pudori] calori; 6.27 quos Riese ] quoque; 6.441 uisendae Heinsius ] uisendam; 7.30 segeti] segetis; 7.62 *concurrere] occurrere; 7.115 ignes Naugerius ] illos; 7.116 anhelatos ] anhelantes; 7.162 *cadit] facit; 7.224 altum] altus; 7.248 ciuit Heinsius ] lenit; 7.375 *spreto] stricto; 7.739 *paciscor] loquendo; 8.117 obstruximus orbem] exponimur orbae; 8.588 *partesque resoluit] pariterque reuellit; 8.820 *spargit] peragit (-at); 8.876 *deerantque] dederatque; 9.8 fando Burman ] tandem; 9.569 *paulum] pauidum; 9.796 *sociosque] sociusque; 10.65 Stygii Heinsius ] timidus; 10.325 dilectu] delicto; 10.423-424 gelidus … tremor] gelidos … artus; 10.590 ipse] ille; 10.733 caeno Merkel ] caelo; 11.361 templi] ponti; 11.368 fulmineos] fulmineus; 11.529 nouiens Merkel ] celsi; 12.510 insani Austri] inani Austri; 12. 581 Phaethontida] Cygneida; 12.625 *Laerteque] Laertiadaeque; 13.235 repono] reposco; 13.244 eadem] etiam; 13.482 quot … cruores] quid … cruoris; 13.694 pectora] uulnera; 13.724 linguis] pinnis; 14.169 *Ithaceque] Ithacique; 14.269 quoque] quoue; 14.804 accedere] excedere; 14.39 ius caeli Muretus ] cuius caelum; 15.715 *columbis] colubris.

In addition to this selection of readings and necessary conjectures, it will also be helpful to look at a selection of the conjectures printed by Tarrant. Although they may not strike all readers as certain or necessary, they nonetheless address difficulties in the transmitted text, and reveal the qualities of the editor’s judgment in operation:

1.345 iuga Slater ] loca; 1.426-427 quaedam perfecta … quaedam modo coepta van Leeuwen ] quaedam modo coepta … quaedam imperfecta; 1.441 numquam letabilis Housman ] et numquam talibus; 1.727 exercuit Postgate ] terruit;

2.278 fractaque Housman ] sacraque;

3.671 toto Shackleton Bailey ] coepit; 3.678 posse ¹ Housman ] esse; 3.716 fremituque Schepper ] trepidumque; 3.724 dereptis Housman ] deiectis;

4.260 nimborum patiens Bentley ] nympahrum (im)patiens; 4.663 Aetnaeo Housman ] aeterno;

5.224 Perseus Bothe ] Phineu; 5.454 mixtae … polentae Kenney ] -ta … -ta;

6.635 Terei Slater ] Tereo;

7.12 nisi Heinsius ] quid; 7.317 medicamina corpus Korn ] ea corporis artus; 7.791 captare Polle ] latrare; 7.865 consortibus Housman ] cum fortibus;

8.53 rogare Reeve ] rogarem; 8.750 silua ² Heinsius ] omnis; 8.829 incensaque Heinsius ] immensaque;

9.711 indetecta Shackleton Bailey ] indecepta;

10.225 ignarus sceleris Madvig ] +inlugubris sceleris+;

11.611 atricolor Heinsius ] unicolor;

12.61 repens Heinsius ] recens; 13.794 palma Siebelis ] pomis;

14.240 franguntque Kenney ] merguntque; 14.245 crede Shackleton Bailey ] uisa; 14.288 manerem Gruyter ] maneret; 14.765 formae deus apta senili Housman ] formas deus aptus in omnes;

15.478 cruore uacent Postgate ] uacent epulis; 15.625 adlegerit Housman ] adiecerit; 15.838 Pylios Heinsius ] similes.

To round out this sample of variants and conjectures, I include a list of the editor’s own conjectures printed in the text: 1.712 posuisse] tenuisse; 2.506 uolucri] pariter; 3.642 persequitur retinens] pro se quisque timet; 3.720 Autonoes ( palmary ] Autonoe; 3.731 derepta] direpta; 5.482 laesa] sparsa; 7.375 at] et; 7.741 male uictor ego en ego fictus] see apparatus for numerous variants; 9.728 me] mihi; 10.257 sed] et. The spelling Porthaoniae (8.542) and Porthaone (9.12) for Parth- of the manuscripts appears to be Slater’s.

Students of the poem will be discussing and debating the editor’s choices, both those presented here in the above selections and numerous others throughout the poem, for a long time to come. There is much here that is new, much that is exciting, and much that will command assent. But inevitably there will be disagreements. The fact that they are so few bears witness to the sharpness and steadiness of the editor’s judgment. The manuscript abbreviations of the OCT are used throughout.

1.92 ligabantur Ω ] legebantur F ⁴ χ. Although ligabantur has the unanimous support of the primary manuscripts, its appropriateness in this context is not at all clear. Helm ( GGA 177 [1915] 531) pointed out that the tense of fixo does not give the proper sequence of events with ligabantur. One would expect that the laws were inscribed on the bronze first, and then the bronze was nailed up in public; the tense of fixo, interpreted strictly, reverses that order. His point is backed up by the coda to the S. C. de Bacchanalibus in which the Senate instructs that its resolution be inscribed on bronze, and then that it be nailed up ( figier) in a public place. One might answer that the participle is proleptic (“which was then nailed up”), but that still leaves the use of the verb ligare for the act of inscribing unexplained. Helm also observed that if legebantur is read, we have an instance of etymological wordplay in lege [90] … legebantur.

1.190 temptanda Ω ] temptata Bern (N ^ac) L ^ac. Tarrant apparently follows Lee, who in his edition of Book 1 (Cambridge 1953) rejected temptata, a reading unanimously accepted by modern editors (“modern” meaning Merkel and after), and printed temptanda. Lee’s defence of temptanda is not, in my judgment, persuasive. He focuses too narrowly on cuncta prius temptanda as an aphorism. If it is to be taken as such, it follows oddly after the pronouncement perdendum est mortale genus (188), which makes it clear that even the consideration of remedial measures is now out of the question. Lee’s objection “that the force of sed after it [ temptata ] is not so good” is answered by Propertius 3.21.5-6. As for Lee’s objection that no attempt has been made to improve human behavior, Jupiter’s decision to walk among mortals in human guise for a purpose other than sex is in itself the most extreme measure that could have been taken. Finally, with regard to transmissional probability, it is much more likely that temptanda is due to the assimilation of temptata to its syntactic environment ( perdendum 188, recidendum 191) than that an orginal temptanda was altered to temptata.

2.83 Scorpion atque aliter curuantem bracchia Cancrum. Riese questioned the genuineness of this line (“interpolatus?”) and his doubt is worth recording. The repetition of the phrase curuantem bracchia in the same metrical position in 83 and 84 with no apparent rhetorical effect; the unusual, in epic poetry, use of aliter in the sense “in a different way”, i.e., in contrast to circuitu longo; and finally, Cancer’s rather tame attribute of stubby arms all suggest that 83 was cobbled together to identify Scorpio by name and to distinguish the long reach of that constellation’s fearsome claws from the shorter reach of Cancer.

2.326 saxum M ^c BG ^3v L] fatum HN ^c B ^2v FG ¹ P factum N ^ac U. Modern editors without exception print saxum. But parallels (9.566, 14.433; Aen. 3.287) suggest that what is being commemorated, Phaethon’s fatum, should be the object of signant, rather than the location of the commemorative words. Fatum is the more expressive word and does not conjure up a picture of the Hesperian Naiads as stonecutters. The typographical device of inscriptional capitals for the text of 327-328 does not provide an argument in favor of saxum.

3.567 remoraminaque M ^2v P] moderaminaque Ω. Of these two variants moderamina is better to suited to ipsa, which suggests that the subject of nocebant is something which would normally have a good effect. Remoramina is the result of the harmonization of the narrative with the eye-witness description of the torrent when it meets with obstructions. 4.388 incesto E ² (M ^ac)(N ^ac)(S)(G^ac)] incerto Ω. With the exception of Anderson, all modern editors print incesto. However, the transformation of Hermaphroditus is not a violation of religious taboo and it does not involve sexual intercourse that is forbidden by law or custom. Defenders of incesto cite infamis … Salmacis (4.285-286) and Pythagoras’s phrase obscenae Salmacis undae (15.319), but neither infamis nor obscenae is weighty enough in its opprobrium to justify the epithet. The prudish variant incesto was concocted to condemn the innocent pleasures of skinny dipping and to contain the threat of sexual abnormality. Incerto, on the other hand, which is defined by neutrumque et utrumque (379), follows naturally after semimarem (381), semiuir (386), and biformis (387). Moreover, if incesto medicamine is read, we are confronted with the scandalous situation in which Hermaphroditus’s divine parents are themselves involved in committing what is incestum in response to their son’s prayer.

8.237 limoso … elice (Auc. de dubiis nominibus GLK V.587)] ramosa … ilice Ω. The attractions of a recherché reading, the agricultural technical term elix, retrieved from the indirect tradition, should be resisted. A “muddy drainage trench” does not provide the vantage point which, as A. Hollis (Oxford [1970)] remarked, is implied by prospexit.

10.584 insidiasque Shackleton Bailey] inuidiamque Ω. The transmitted text gives excellent sense: Hippomenes “fears his feelings of jealousy”. Why? Because, unlike other lovers who feel the pangs of jealousy, Hippomenes must risk his life to win Atalanta ( mors pretium tardis 572, tanta pericula 576). His inuidia is expressed in the clause ne … optat (583); in wishing that none of the other suitors runs faster than Atalanta, Hippomenes realizes the inevitable consequence of his jealousy, that he must run for his life and his love. A jealousy that may prove fatal is reason enough to be afraid. With sed in 584 Hippomenes emerges from his emotional turmoil and is resolved to act. Some editors resorted to personification, Inuidiam, a desperate measure.

11.627 Herculea Trachine. The local ablative Herculea Trachine, construed with adeant, although accepted by all editors, is a solecism; “to go to Alcyone at Herculean Trachis” (Hill) is the English, not the Latin, idiom. The accusative, Herculeam Trachina, should be read. The accusative of the adjective is well attested by the manuscripts: herculeam M ^c N ^c FGL. And although there is no note in the apparatus for Trachine, M’s variant thracena, reported by Anderson, offers good evidence for the accusative. It is also worth mentioning that Planudes read Trachina.

12.278 at illud Ω. The adversative at seems harsh in this context since illud is merely a continuation of the relative quod, which refers to ferrum (276). Ovid regularly uses at with a form of ille at line-end when he wants to shift attention from one subject to another or to draw a contrast. Here there is no such shift or contrast. Magnus reports et from the Rome edition of 1471. Whether it is a conjecture or a misprint, it gives excellent sense. A comma, rather than a semi-colon, after demittit serves the construction better, and a comma after eduxit would be helpful.

12.492-493 caecamque … manum Ω ] caecaque … manu N ^ac. Caeneus has just buried his sword up to the hilt in Latreus’s forequarters. It seems odd to read that Caeneus then moved his hand into the centaur’s guts and inflicted a wound within a wound. The reading of N before correction, caecaque … manu gives, I think, better sense: after driving the sword in up to the hilt, Caeneus then moved and twisted it ( ensem is the object of mouit [492] and uersauit [493]) into the creature’s vitals “with a blind hand”, i.e., a hand that could not be guided by the eye as it wielded the blade into Latreus’s vitals.

13.243 sic N ^c φ ] si (M ^ac)(N ^ac)ὑπ ^ac). (Ulysses is responding to Ajax’s criticism of his night mission with Diomedes.) The words si tamen have not been correctly understood and deserve a second look. It is not enough for Ulysses to say in 241-242 that it counts for something to have been chosen by Diomedes for the night mission, since Ajax mocked Ulysses’ role in it (98-106) and the mere assertion of merit does not answer Ajax’s scornful dismissal. Ulysses must back up his claim. He does so by continuing the sentence that begins est aliquid with a clause that supports his assertion. The phrase si tamen, meaning “in spite of what has been said” and, anticipating skepticism on the part of his audience, introduces Ulysses’ grounds for making such an assertion. First, he claims, in response to Ajax’s belittling comments (13.15, 100), that he was taking a real risk, not shielding himself under cover of darkness, when he undertook the night mission; et … periclo“scorning the dangers of night and the enemy as well” ( et here adverbial, meaning “as well as Diomedes”, whom Ajax does recognize as a courageous fighter) suggests that encountering the enemy at night requires more courage than facing them in the light of day. Second, he claims, again in response to Ajax’s criticisms (13.98, 100), that, without Diomedes’ help, he killed a man who was a real threat to the Greeks. Translate: “It counts for something to be the one man alone chosen by Diomedes out of so many thousands of Greeks (and in fact the lot was not commanding me to go), if, in spite of what has been said, scorning the dangers of the night and the enemy as well, I kill Dolon … ” nec … iubebat is a parenthesis.

14.305 illum/illis Ω ] ipsi e W. The editor prints illum. The demonstrative is superfluous in this expression of reciprocal action; ipsi, on the other hand, is the right word, as the parallels in Bömer show, to emphasize the notion of mutual feeling and response.

14.434 Camenae μ ^ac)] coloni Ω. The Camenae make a very abrupt appearance here, without so much as a relative clause explaining how they are involved in the story or why they take an interest in the fate of Canens. The etymological connection ( Canens — carmen — Camena is insufficient to fill this gap in the narrative. At 15.482 the Camenae are completely at home in the company of Numa and his wife Egeria. The suggestion that coloni may be an interpolation due to ueteres coloni at 6.318 may cast doubt on the word. But that doubt must be weighed against the possibility that Camenae is an interpolation due to nympha … Camenis at 15.482.

14.657 + tantoque potentior +] tantoque peritior N ^c U G ¹ P. The variant tantoque peritior has been undeservedly neglected by editors. It is consistent with Pomona’s surpassing skill as an orchardist (14.623-627), and it gives excellent sense. When Vertumnus, disguised as an old woman, sees the proof of Pomona’s special abilty, he compliments her using a colloquial expression, tanto peritior, which accounts for laudatae in the following line. The agreement of U and P in reading peritior also suggests that it merits serious consideration.

14.466-467 alta cremata est Ilion Ω ] Ilios φ. The neuter Ilion, with alta and cremata, is inconsistent with Ovid’s practice at Her. 1.48, and Ars. am. 1.363, in both of which the less well attested variant Ilion is also transmitted. Given Ovid’s practice and the Homeric epithet phrase ( Il. 13.773) on which the Latin is modeled, Ilios has the better claim to be correct. Palmer’s note on Her. 16.49 (Oxford 1898) is more helpful here than Bömer’s. While on the subject of orthography, Lachmann’s Thrace for Thracia Ω (6.435) should at least be mentioned, since it agrees with the form used by Ovid elsewhere. Lachmann observes, “neque Thraciam ullus poeta dixit praeter Lucanum” (p.278, on Lucretius 4.30-31). If Ovid was so fastidiously hellenizing in his use of Greek forms as he is represented to be in this edition, then the possibility that he wrote the Greek genitive Androgeoque for Androgeique Ω cannot be ruled out (7.458); cf. Aen. 6.20. Overall, considerable care and attention have been given to the orthography of the text, which is a marked improvement over what is found in Magnus, Ehwald, and Anderson.

Punctuation.

1.392: Ehwald’s punctuation of pia sunt … suadent as a parenthesis is worth mentioning. The parenthesis functions as a suspenseful postponement of the correct interpretation of the oracle, which then produces a better disjunction, aut fallax est sollertia nobis aut (pia … suadent) magna parens terra est.

2. 436-437 (Callisto is putting up a fight against Jupiter.) The punctuation adopted by all modern editors, a question mark after poterat in 437, with superum, an epithet of aethera, beginning a new sentence, spoils the sense. (Bömer notes that superum aethera is an unparalleled phrase.) The question mark belongs after superum, which is a partitive genitive dependent on both interrogatives, quem (436) and quis (437): “But whom of the gods could a girl overcome, or who of the gods could overcome Jupiter?” Ovid is clearly sympathetic to Callisto’s helplessness against such a powerful rapist and cleverly amplifies the magnitude of her disadvantage by pointing out that she is not only matched against a god but against a god whom no other god could defeat. To ask, in this context, who could whip Jupiter in a fight, without specifying that Jupiter’s opponent is a god, is inept; and to imply that the girl couldn’t whip anybody in a fight is scornfully dismissive and inappropriate to the poet’s concern for a victim who is overpowered by an invincible divinity. Note the wordplay: even beings who are superi cannot superare Jupiter.

6.8-9 pater huic Colophonius Idmon / Phocaico bibulas tingebat murice lanas. This one sentence is, I believe, two. Punctuate with a semi-colon after Idmon, “her father was Idmon of Colophon”; the next sentence gives the father’s occupation. Otherwise the dative would most naturally be taken to mean that the father dyed wool “for her”. The general assumption that huic here = huius needs to be reconsidered.

8.631: Ehwald’s semi-colon after sed pia provides the expected contrast to parua quidem. In addition, pia referring to the home of Philemon and Baucis, helps to explain the emphatic illa … illa … casa (633).

13.120: quid uerbis opus est is better punctuated with Ehwald as a parenthesis since denique functions best as the introduction to Ajax’s summation, spectemur agendo, and the impatient question regularly occurs without any introductory word or with sed.

14.159-160 The period at the end of 159 means that qui in 160 is to be taken as the connective relative. This punctuation, however, results in a violation of standard word order: the connective relative comes first in its sentence except when preceded by one other word, usually a preposition. Since qui in 160 is postponed to fourth position, it is very unlikely that it functions as the connective relative. A comma is needed at the end of 159. However, qui, which is found in one or more of the twelfth century manuscripts, is one variant among several; of all the variants N’s sub, defended by Helm ( GGA 177 [1915] 549), seems right; Achaemenides was abandoned “under the heart of Aetna’s crags”, i.e., deep inside the Cyclops’s cave at the foot of the mountain.

The apparatus criticus is a model of clarity and concision; in fact, it is a miraculum worthy of the poet’s art, the transformation of a vast body of seemingly chaotic textual data into a well-ordered cosmos where one can chart the course of the transmission and study the evidence upon which countless textual decisions are based. Those qualities of clarity and concision are the result of an exemplary Editionstechnik that eschewed the accumulation of unnecessary information, such as the names of scholars who support various readings, and of information which is more conveniently presented elsewhere, such as variations in spelling and morphology. That material has been painstakingly gathered in the appendix orthographica et morphologica (483-501), an important contribution in itself to our understanding of this neglected field.

The text and apparatus have been produced with remarkable accuracy. I noticed only a handful of slips. 8.8 Alcathoi for Alcathoe is more than a spelling variant and should reported in the apparatus, not the orthographical appendix. 11.49: the adoption of the variant habuisse requires the change of the nominatives Naides and Dryades to the accusative, Naidas and Dryadas, but the nominatives remain in the text. 13. 843-844: something is missing in the note. The parenthesis ( nam … Iouem) is attributed to Magnus, but is found earlier than his edition. What Magnus suggested is that in caelo belongs in the parenthesis. In three places the note in the apparatus indicates that the wrong word has been printed in the text: 8.576, gerat/gerit; 13.460, ferrem/vellem; 14.272, omnia/omina. The apparatus is silent on the omission of 1.8 in Bern, and the variant coegit (1.33) in the same. There is no note on the variant potiremur at 13.130. According to the note at 7. 657, the line should be bracketed; apparently it has been promoted to the ranks of suspicious verses. There is a misprint at 10.380, intellige.

A full Index Nominum rounds out the volume. It contains at least one new entry, Lindius; cf. 13.684.

I close with a few conjectures.

3.241 circumfert tacitos tamquam sua bracchia uultus Ω. Actaeon the stag is the subject of circumfert. Neither circumfert nor tacitos is appropriate to bracchia, which is regularly used with tendere to describe the posture of the suppliant, whether or not the suppliant actually has arms to extend. Even the most aggressive enthusiast of zeugma will not find a way to interpret this line as a characteristically Ovidian analogy that relates an anatomical feature of a transformed being to the corresponding feature of its original self. One need only look at a translation to see the glaring incongruity of tacitos … uultus and bracchia, an incongruity only heightened by sua. Most importantly, the point is not that silent facial expressions are a substitute for the suppliant’s extension of the arms; rather they are a substitute for speech. In the immediately preceding lines (237-240) we are given a description of the stag’s inability to communicate. (Cf. Fast. 2.613, uoltu pro uerbis precatur) The word that fits the analogy and fits both circumfert and tacitos is lumina : “he casts about silent expressions as if the look of his own eyes”. Similarly Callisto the bear keeps her eyes fixed on her son Arcas as if to show that she recognizes him (2.501-504). Vultus and lumina/oculi form a natural pair; see 14.840; Her. 16.148; Trist. 5.4.39; Vergil, Aen. 6.156, 6.862.

3.642 persequitur retinens Tarrant] pro se quisque timet Ω. Although the transmitted text presents serious problems of interpretation, not least of which is the fact that the mutineers appear not to fear for themselves, nevertheless quisque itself strikes me as genuine because the distributive pars … pars follows in 643. I would suggest praedae for pro se, and punctuate the sentence as a parenthesis which explains the behavior of the crew: ( praedae quisque timet (“each one is afraid for his plunder”), i.e., the kidnapped Bacchus, when they realize that Acoetes is steering the ship to Naxos. There are earlier references to the kidnapped god as plunder at 3.606 and 3.620. This use of the dative with timere in epic poetry has good parallels at Aen. 2.130 and 11.550. quis te furor — (641) will be an aposiopesis, more an expression of anger than a question.

4.341 ut puer (F ⁴ L) et uacuis (EM ^c) ut Heinsius] scilicet ut uacuus et Ω. If ut inobseruatus in this line represents Hermaphroditus’s false assumption that he is not being watched, then it is incompatible with ut puer, which cannot represent what the boy is thinking as he prepares to dip his feet in the water, but rather is appropriate to the narrator’s view of the action. If, on the other hand, ut puer is understood as the narrator’s comment, then it is incompatible with ut inobseruatus, which the narrator knows to be false. We cannot have the following shift in perspective; “seeing that he is a boy” (narrator), and “seeing that [he thinks] that he is not being watched” (narrator’s representation of what the boy is thinking). Moreover, this use of causal ut to introduce a supposition that is false is a peculiar one since it is normally used to introduce a consideration or reason that supports an assertion. Because of these difficulties I would prefer to retain scilicet and read ( scilicet it uacuis ut inobseruatus in herbis) (“naturally he goes about as though unwatched in the deserted grass”). This parenthesis is the narrator’s suggestive hint at what Hermaphroditus might do in such surroundings while under the mistaken impression that he is alone, and at what Salmacis might hope to see from her secret vantage. The repetition of it, once in a parenthesis and once in the narrative (342), is not offensive. 4.414 tectaque Ω. Caves are the natural habitat of bats (cf. Homer, Od. 24.6). Read antraque. The transmitted tectaque is due to harmonization with a particular detail of the narrative, latitant per tecta (404). 8.279 at non impune feremus Ω. With these words Diana vows to avenge Oeneus’s slight to her divinity. The difficulty is that impune ferre is the standard Latin idiom for “to get away with” a crime or offence, and Ovid always uses it in that sense. A different idiom is required here, impune sinere“to let (someone) go unpunished”. Read non impune sinemus, “we will not let him go unpunished”. Cf. 5.119, 6.4, 11.67.

9.74 + reduxi + Ω. Merkel’s emendation reclusi, which he understood in the sense to “lay open” or “open up” a part of the body, was accepted by Magnus, Ehwald, and Anderson. Tarrant is right to reject it because it doesn’t work in this context. It is one thing to say recludere pectus / iugulum / uulnus, quite another to say recludere echidnam. The apparatus records several attempts to replace reduxi with a verb meaning “to burn” in order to include a reference to Iolaus’s cauterization of the hydra’s severed necks. Merkel showed that these attempts are misguided when he wrote “cum tamen non domita perusta, sed perusta domita hydra fuerit”. We’re looking for a word to express the coup de grâce. The adjective ramosam (73) and the participle crescentem (74), creating a picture of vegetative growth, suggest that Hercules may have repeated the verb he used at the end of 71 ( recisum); he will then have said domitamque recidi.

9.365 + loton + Ω. The transmitted text is rightly obelized by the editor. It is in all probability a gloss that displaced the original reading. Using 1.556 as a parallel ( oscula dat ligno, refugit tamen oscula lignum), I suggest that the original reading in 9.365 was lignum : ostendi lignum. tepido dant oscula ligno. This type of repetition is an Ovidian mannerism.

9.779 quod non ego +punior ecce+ Ω. This is the second of two quod -clauses in Telethusa’s prayer of thanksgiving to Isis. In the first clause ( quod uidet haec lucem) she thanks Isis for the life of her daughter Iphis. When Telethusa was pregnant with Iphis, her husband Ligdus had instructed her to kill the child if she gave birth to a girl. But, after Isis appeared to her in a dream and told her to raise the child, whichever sex it was, Telethusa raised her daughter disguised as a boy. The second quod -clause will yield satisfactory sense if we read quod ego non funeris auctor. The conjecture funeris auctor, which is fairly close to the transmitted text, is based on the language used by parents who contemplate being the agent, in one way or another, of their child’s death: Althaea and Meleager, mortis auctor 8.493; Sol and Phaethon, funesti muneris auctor 2.88; Dido, Aeneas, and their unborn child, funeris auctor, Her. 7.136. Telethusa will then be thanking Isis “for the fact that this girl sees the light of day and that I am not an agent of death” (which she would be if she had followed her husband’s order). For the omission of sum in a subordinate clause, coupled with ego used in contrast to another pronoun, see Her. 6.148 and 10.143.

10.637 quod facit G] quid facit Ω. quid facit, printed by Magnus, Ehwald, and Anderson, was selected by Housman as a specimen of what the art of conservative criticism can do ( Manilius 1.xlviii). The adoption of G’s quod is an expedient which removes the problem of the indicative in an indirect question, although one wonders why Ovid didn’t write the line using the standard expression quid faciat. The relative, however, does not solve the problem of sense; Atalanta is experiencing an emotion that is new to her, and quod facit doesn’t seem to fit the general idea that she does not understand what she is feeling. If we read quid ferat ignorans“not knowing what she is undergoing”, then the poet will be emphasizing the novelty and strangeness of what Atalanta feels. This is an internal transformation. 11.153 iactat sua carmina Ω. The sense is questionable and the repetition carmina … carmen (154) is rhetorically flat. The editor observes of carmina“nescio an sanum”. Pan is a dancer as well as a musician. If we read iactat sua corpora, we have the traditional picture of Pan in performance.

11.363-364 palus … quam … paludem Ω. The repetition of the antecedent ( palus) of a relative pronoun in the relative clause ( quam … paludem) is a feature of Latin prose style that is very much at home in Caesar’s Bellum Gallicum but not at all in the Met.. Read lacuna for palus in 363, iuncta lacuna huic est : “Next to this grove was a hollow … which overflowing sea water made a marsh.” Cf. 8.335-337. Since, however, elision with huic is rare in the Met., it may be preferable to transpose huic and est : iuncta lacuna est huic densis obsessa salictis.

11.610 ebeno Ω. Ebeno is taken with torus to mean “a couch made of ebony”. However, this use of the bare ablative to indicate the material of which something is made is unusual since Ovid regularly uses e/ex with the ablative as an attributive modifier to describe an object’s material composition. Cf., for example, 4.638, ex auro ramos, ex auro poma. I suggest torus ex ebeno for torus est ebeno. Another, less attractive, possibility, if est is to be preserved, is the genitive, ebeni, but Ovid rarely uses this construction (cf. 11.515 tegmine cerae).

13.922 ducentia Ω. The note records Reeve’s suspicion that ducentia may be an unconscious repetition of the stem of the immediately preceding ducebam. If so, then fallacia is a good possibility; cf. Ars am. 2.189, fallacia retia.

14.166 iam suus et Ω. It is puzzling that the reference to Achaemenides’ psychological condition is sandwiched between descriptions of his sartorial condition ( hirsutus amictu 165; conserto tegmine 166), especially so since this awkward arrangement is emphasized by the anaphora and asyndeton of iam. For iam suus et read usus iam; usus here means “wearing”.

15.104 + deorum + Ω. As the editor notes, deorum was interpolated from 1.32. All of the conjectures in the apparatus provide a genitive plural dependent on uictibus, even though the genitive may be nothing more than a false lead due to the interpolated deorum. An epithet for uictibus is also a possibility; cruentis has the advantage of giving us a picture of the bloody meal and of conveying the moral pollution attendant upon it.

15.122 demum Ω. The editor observes of demum“plerisque suspectum”; and rightly so since it is usually interpreted as if it meant no more than quidem or uero emphasizing immemor. Read operum, “unmindful of work in the fields “, which requires the assistance of the ox; opera, the general term for fieldwork performed by the ox (cf. Fast. 1.83 and 4.336), is then specified in 123-126. The adjective immemor characterizes the farmer as an ungrateful friend who repays the faithful services of his helper in the fields with slaughter.

Tarrant’s OCT of the Metamorphoses will, in my judgment, be recognized as one of those rare critical editions which effects a permanent improvement of the text and sets a very high standard against which future editions of the poem will be measured. It was well worth the wait.