81-86. Ovid similarly calls Vestalis as his witness at vii 3-4 'aspicis en praesens quali iaceamus in aruo, / nec me testis eris falsa solere queri'.

81. INCOMMODA. The word is not found elsewhere in Ovid, and is not used in verse, except for satire (Hor AP 169; Juvenal XIII 21). It is particularly common in Caesar.

81. CAELI = 'climate', as commonly (Tr III iii 7, Prop II xxviii 5, Cic Att XI xxii 2).

82. QVAM VICINO TERREAR HOSTE ROGA. An imitation of Tib I i 3 'quem labor assiduus uicino terreat hoste'.

83. SINTNE LITAE TENVES SERPENTIS FELLE SAGITTAE. Similar descriptions of poisoned arrows at Tr IV i 77 'imbuta ... tela uenenis', Tr IV i 84, Tr III x 64, Tr V vii 16 'tela ... uipereo lurida felle', EP I ii 16 'omnia uipereo spicula felle linunt', EP III i 26, and EP III iii 106.

84. FIAT AN HVMANVM VICTIMA DIRA CAPVT. Human sacrifice similarly mentioned at Tr IV iv 61-62 'illi quos audis hominum gaudere cruore, / paene sub eiusdem sideris axe iacent'.

85. MENTIAR. Professor J. N. Grant points out to me the asyndeton following quaere ... sintne. Compare the similar problem at iv 31-32.

85. AN COEAT DVRATVS FRIGORE PONTVS. Similar wording at vii 7 'ipse uides certe glacie concrescere Pontum', Tr II 196 'maris astricto quae coit unda gelu', and Tr III x 37.

86. IVGERA MVLTA FRETI. According to TLL VII.2 629 7-8 this is the unique instance of iugerum being applied to water. The transferred sense is natural enough in view of the poets' application to the sea of such words as campus and arua.

89. NON SVMVS ... ODIO. Basically a prose use; but compare Met II 438 'huic odio nemus est', Fast VI 558, EP II i 4 'iam minus hic odio est quam fuit ante locus', and Ecl VIII 33 'tibi est odio mea fistula'.

Owen's second edition has the misprint 'nec sumus hic odio', reproduced by Wheeler. The error was induced by nec at the start of the pentameter.

90. NEC CVM FORTVNA MENS QVOQVE VERSA MEA EST. For Ovid's use of syllepsis, see at vi 16 spem nostram terras deseruitque simul (p 234). For the sentiment of this line, compare Sen Med 176 'Fortuna opes auferre, non animum potest', where Costa cites Accius 619-20 Ribbeck² 'nam si a me regnum Fortuna atque opes / eripere quiuit, at uirtutem non quiit', Sen Ben IV 10 5, Sen Ep XXXVI 6, and Euripides fr. 1066 Nauck.

91. ILLA QVIES ANIMO. Animo is locative; or perhaps in should be supplied from the following line: for the joining of a noun with a following preposition already with a complement, see Clausen on Persius I 131 'abaco numeros et secto in puluere metas'. I read animo (found in one of Heinsius' Vatican manuscripts) because of the parallel structure it gives with the following in ore, but ANIMI (BCMFHILT) is possible enough: OLD quies 7 cites quies animi at Celsus III 18 5.

91. QVAM TV LAVDARE SOLEBAS. The same phrase at Her XV 193 'haec sunt illa [sc pectora], Phaon, quae tu laudare solebas'. For the persistence of Ovid's old habits, compare EP I x 29-30 (he remains a moderate drinker, as formerly).

93-94. SIC EGO SVM LONGE, SIC HIC, VBI BARBARVS HOSTIS / VT FERA PLVS VALEANT LEGIBVS ARMA, FACIT is clearly corrupt, as will be seen from Wheeler's 'Such is my bearing in this far land, where the barbarian foe causes cruel arms to have more power than law' and André's 'Je vis au loin, ici, où un ennemi barbare donne aux armes cruelles plus de force qu'aux lois'. Merkel ejected the distich, which seems the best solution; it is not necessary to the poem's structure, and the iterated facit ut in unrelated clauses at 94 and 97 is suspicious. Also, as Professor R. J. Tarrant notes, the ut in 94 makes one expect that ut in 95 will be correlative, when it in fact continues the thought of 93 (or rather of 91-92, after 93-94 are excised).

Heinsius thought 93 alone to be suspect; if so, the meaning lying behind the text is probably something like 'What I once was at Rome, I still am here'.

93-94. HIC, VBI BARBARVS HOSTIS, / VT FERA PLVS VALEANT LEGIBVS ARMA FACIT. Similar statements at Tr V vii 47-48 'non metuunt leges, sed cedit uiribus aequum, / uictaque pugnaci iura sub ense iacent' and Tr V x 43-44; see also Otto lex 3.

93. BARBARVS HOSTIS. The same phrase at Tr III x 54, Tr IV i 82, and EP II vii 70.

95. RE ... NVLLA MHIL REM NVLLAM BCFT. The verb queri can take a direct object, or be constructed with de + ablative, but not both; this would in effect give the verb two objects. Re ... nulla removes this difficulty and is obviously prone to corruption, the true object de nobis being postponed to the following line.

96. FEMINA ... VIRVE PVERVE = 'anyone'; compare Tr III vii 29-30 'pone, Perilla, metum: tantummodo femina nulla / neue uir a scriptis discat amare tuis', and Ovid's use of femina uirque 'everyone' at Met VI 314-15 'femina uirque timent cultuque impensius omnes ... uenerantur numina', RA 814, Tr I iii 23, and Tr II 6. The repeated u in uirue would not have offended the Romans: compare for instance Tr III vii 30 'neue uir', Am I viii 97 'uiri uideat toto uestigia lecto', and Met XII 204 'poteratque uiri uox illa uideri'; conscious alliteration at Am III vii 59 'uiuosque uirosque' and Met XIII 386 'inuictumque uirum uicit'.

98. HAEC QVONIAM TELLVS TESTIFICANDA MIHI EST. Similar phrasing at Ibis 27-28 (of Augustus) 'faciet quoque forsitan idem / terra sit ut propior testificanda mihi'.

100. RESPECTV ... SVI. 'Out of consideration for themselves'. Respectus elsewhere in Ovid only at Tr I iii 99-100 (of his wife after his departure) '[narratur ...] uoluisse mali [Madvig: mori codd] moriendo ponere sensus, / respectu tamen non periisse mei'. Respectus is found in Phaedrus, Martial, and Juvenal, but not in Virgil, Horace, or Propertius.

101. NEC MIHI CREDIDERIS in its absolute use here seems colloquial: elsewhere Ovid uses nec ... credideris to introduce a dependent clause (Tr V xiv 43; EP I viii 29).

101. EXTANT DECRETA QVIBVS NOS / LAVDAT ET IMMVNES PVBLICA CERA FACIT. The same honour described in greater detail at xiv 51-56.

101. EXTANT ('there exist') is somewhat more forceful than the nearly equivalent sunt: compare xiv 44 'extat adhuc nemo saucius ore meo', Cic Planc 2 'uideo ... hoc in numero neminem ... cuius non extet in me summum meritum', and Cic Diu I 71.

102. PVBLICA CERA = tabulae publicae, 'public records', for which compare Cic Arch 8 & Fl 40, and Livy XXVI 36 11. The same metonymy at Val Max II x 1, where tabulae and cera are used as synonyms, and at Hor Ep I vi 62 'Caerite cera', where commentators cite Aulus Gellius' mention of tabulae Caerites (XVI 13).

103. QVAE R. J. Tarrant HAEC L, probante Heinsio ET BCMFHIT. Quae connects with idem in the following line and provides a more satisfactory sense than et, which would make the sentence mean that Ovid did not consider the decrees something to boast of. Quae quamquam is preferable to haec quamquam since it connects better with the preceding line and is obviously more prone to corruption; but for a similar corruption of haec compare Prop II xxiii 1 'fuit indocti haec [uar et] semita uulgi'. For quae Professor Tarrant cites EP III v 9-10 'quae quamquam lingua mihi sunt properante per horas / lecta satis multas, pauca fuisse queror' and EP III viii 23-24 'quae quamquam misisse pudet ... tu tamen haec quaeso consule missa boni'.

103. QVAMQVAM ... SIT G QVAMQVAM ... EST BCMFHILT. For the subjunctive Luck compares Met XIV 465 'admonitu quamquam luctus renouentur amari' and Met XV 244-45 'quae [sc elementa] quamquam spatio distent, tamen omnia fiunt / ex ipsis'; in the first passage a few manuscripts and in the second the majority offer the indicative. Ovid usually has the indicative following quamquam; but sit should be taken as the correct reading here in view of G's early date.

105. NEC PIETAS IGNOTA MEA EST. At xiii 19-38 Ovid describes an instance of his pietas, the reciting to the Getes of a poem in Getic on Tiberius.

105-10. The figures of the imperial family had been a gift of Cotta Maximus, for which EP II viii was a letter of thanks. For a discussion of Ovid's treatment of the imperial family, particularly in the poems of exile, see K. Scott "Emperor Worship in Ovid", TAPA LXI [1930] 43-69.

106. CAESARIS. Augustus, as is made clear by the next line.

107. NATVSQVE PIVS. Tiberius; see at viii 63 auum (p 277). For Tiberius' piety to Augustus' memory compare Tac Ann IV 37 4 (AD 25; Tiberius speaking) 'cum diuus Augustus sibi atque urbi Romae templum apud Pergamum sisti non prohibuisset, qui omnia facta dictaque eius uice legis obseruem, placitum iam exemplum ... secutus sum'.

107. CONIVNXQVE SACERDOS. Livia, priestess of the deified Augustus; Germanicus was his flamen. For the language compare Vell Pat II 75 3 'Liuia ... genere, probitate, forma Romanarum eminentissima, quam postea coniugem Augusti uidimus, quam transgressi ad deos sacerdotem ac filiam'.

108. FACTO ... DEO. See at viii 63 quem uirtus addidit astris (p 277).

109. VTERQVE NEPOTVM. Germanicus and Drusus.

111. PRECANTIA VERBA = preces. The same phrase at Met VI 164, IX 159, and XIV 365.

112. EOO ... AB ORBE. The same phrase at Fast III 466 & V 557.

113-14. Williams suggested deleting this distich: 'The distance between Tota and Pontica terra, the use of licet=if, and Pontica terra immediately followed by Pontica tellus, point to an interpolation'.

The hyperbaton of tota ... Pontica terra seems standard enough. Wheeler translates licet quaeras as 'you are free to inquire', which may be right; however, the phrase does indeed seem awkward, and licet may be an intrusive gloss that has displaced uelim: compare Her IV 18 'fama—uelim quaeras—crimine nostra uacat'. The repetition of Pontica terra and Pontica ... tellus is a very strong argument for deleting one of the two distichs. However, 115-16 seems more likely to be the interpolation in view of the difficulties discussed in the next note.

115. ORA. Ehwald (KB 65) read ARA (B), citing Dessau ILS 154 14-15 'ara(m) numini Augusto pecunia nostra faciendam curauimus; ludos / ex idibus Augustis diebus sex p(ecunia) n(ostra) faciendos curauimus'; but the ara and ludi are clearly separate items in the inscription, which does not support the phrasing ara natalem ludis celebrare.

Even with ora, 115-16 read rather oddly: the notion of an individual conducting ludi is strange, and the singular dei seems rather vague after the collective his of 111. If the distich is excised (as Professor R. J. Tarrant suggests) 113-14 round out the paragraph that began with 105 (note the correspondence of uidet hospita terra in 105 with testis Pontica terra in 114), and 117 introduces hospites as a second class of witnesses.

118. LONGA. Not 'distant' (Wheeler) but 'long'; compare Met XIII 407 'longus in angustum qua clauditur Hellespontus'. Longus meaning 'distant' is extremely rare: OLD longus 6 cites only Silius VI 628 'remeans longis ... oris' and ps-Quintilian Decl 320 6 'longas terras ... peragraui' (Lewis and Short add Justinus 18 1 'longa a domo militia'). The normal Latin words for 'distant' were longinquus and longe (ancestor of French loin).

119. IS in its various forms occurs only seven times in EP IV: the other occurrences are of feminine singular ea at i 17, viii 27 & xiv 11, of eius at xv 6 (its only occurrence in the Ex Ponto), of accusative id at i 19, and of accusative neuter plural ea at x 35.

The elegiac poets avoided the use of is, preferring hic, ille, and iste. The singular nominative forms were the only ones used relatively freely by Ovid (about forty instances of each); Tibullus and Propertius avoided even these (Platnauer 116; Axelson 70-71).

119. QVO LAEVVS FVERAT SVB PRAESIDE PONTVS. See at 75 praefuit his ... locis modo Flaccus (p 308).

119. LAEVVS ... PONTVS = Euxini litora laeua (Tr IV i 60). A similar brachylogy at EP I iv 31 'iunctior Haemonia est Ponto quam Roma sinistro [Burman: sit Histro codd]'.

119. PRAESIDE. This seems to be the first instance of praeses 'governor' in Latin. It is found in prose from Tacitus and Suetonius on: Trajan even uses it in his official correspondence (Pliny Ep X xliv).

119. FVERAT. See at vi 12 nec fueram tanti (p 230).

121. AVDIERIT. Probably a perfect subjunctive 'may have heard', although possibly an epistolary future perfect indicative ('when you receive this, your brother will perhaps [forsitan] have heard'). For the perfect subjunctive compare Met X 560-62 'forsitan audieris aliquam certamine cursus / ueloces superasse uiros'.

121. FORTVNA EST IMPAR ANIMO. Similar phrasing at Tr V v 46-47 (on his wife's birthday) 'at non sunt ista gaudia nata die, / sed labor et curae fortunaque moribus impar'; but note the different sense of fortuna.

121. FORTVNA. 'My means' (Wheeler). The sense is rare but classical; OLD fortuna 12 cites among other passages Cic Fam XIV 4 2 'periculum fortunarum ['possessions'] et capitis sui' and Caes BG V 43 4.

122. CARPO ... OPES. For the sense of carpo see at viii 32 carpsit opes ... meas (p 266).

126. ILLVM CMFHILTB² ILLI B¹. Either accusative or dative would be acceptable enough with latere. The earliest instances from verse given by TLL VII.2 997 49 are Lucretius III 280 for the dative and Aen I 130 for the accusative. I retain the accusative because it is the reading of most manuscripts, including B's close relative C. There are similar variants involving the object of latere at Fast V 361: the accusative given by most manuscripts is generally read in preference to the dative.

127-29. TV ... TV. For the anaphora of tu in hymns or solemn prayer, see the passages collected by Nisbet and Hubbard at Hor Carm I x 9 and by Tarrant at Sen Ag 311.

127. SVPERIS ASCITE. Asciscere is generally used of admission to the citizenship or to the Senate: for parallels to the metaphorical use here, see Tarrant at Sen Ag 812-13 'tuus ille bis seno meruit labore / adlegi caelo magnus Alcides'.

128. Causal VT ['ex ueteribus' Naugerius] seems an appropriate correction for the manuscripts' lame ET.

129-30. NOSTRAS ... PRECES. The hyperbaton adds elevation and dignity to the prayer.

129-30. INTER CONVEXA ... SIDERA = inter sidera conuexi caeli; the hypallage adds further to the elevation of the passage. For conuexa compare Festus (58 Muller; 51 Lindsay) 'conuexum est ex omni parte declinatum, qualis est natura caeli, quod ex omni parte ad terram uersum declinatum est', Met I 26 'ignes conuexi uis et sine pondere caeli', Ecl IV 50, and Cic Arat 560 (314). In particular compare Aen I 607-8, which Ovid is clearly imitating: 'dum montibus umbrae / lustrabunt, conuexa polus dum sidera pascet'. There is some question as to whether conuexa should there be taken with sidera, or as the object of lustrabunt: Ovid clearly took it with sidera.

130. SOLLICITO QVAS DAMVS ORE PRECES. For the general wording compare Tr III viii 20 'tum quoque sollicita mente rogandus erit' and EP III i 148 'nil nisi sollicitae sint tua uerba preces': for sollicito ... ore compare sollicita uoce at Met X 639 & XIV 706.

131. PERVENIANT ISTVC. Compare EP II ii 95 'si tamen haec audis et uox mea peruenit istuc [=Romam]'.

131-32. CARMINA ... QVAE DE TE MISI CAELITE FACTA NOVO. Ovid also mentions his poems on Augustus' apotheosis at vi 17-18, viii 63-64 & xiii 25-26.

133-34. NEC TV / IMMERITO NOMEN MITE PARENTIS HABES. 'Et ce n'est pas sans raison que tu portes le doux nom de Père' (André) must be correct as against Wheeler's 'for not undeservedly hast thou the gracious name of "Father"', since nec, although it can mean et ... non or sed ... non, cannot mean nam ... non; the proof of this is the frequent occurrence of neque enim.

The litotes non (haud, nec) immerito is common enough in Latin: see the many examples at TLL VII.1 457 26 ff. But in the four instances given of nec immerito, it never serves to introduce a new phrase as here. At Plautus St 28 'decet neque id immerito eueniet' it introduces a second verb which amplifies the preceding one, while it modifies preceding verbs at Ter Ad 615 'tanta nunc suspicio de me incidit neque ea immerito', Val Max IV vii 1 'inimicus patriae fuisse Ti. Gracchus existimatus est, nec immerito, quia potentiam suam saluti eius praetulerat', and Quintilian X i 104 'habet amatores—nec immerito—Cremuti libertas'. One would expect a clause of causation to follow auguror his igitur flecti tua numina, and I think it possible that Ovid wrote NAM TV / E MERITO (Professor C. P. Jones suggests EX MERITO). Both the corruption from e merito and the subsequent interpolation of nec would be easy enough. For e(x) merito, compare vii 16 'contigit ex merito qui tibi nuper honor'.

133. NEC TV. The elegiac poets admitted a monosyllabic ending to the hexameter if it was preceded by another monosyllable closely linked to it in sense: see Platnauer 13. For true monosyllabic endings, see at ii 47 Aonius fons.

134. NOMEN MITE PARENTIS = nomen parentis, quod significat te mitem esse. At Tr I i 73 and EP II viii 51 members of the imperial family are called mitissima numina. There is another instance of hypallage with nomen mite (a different sense of mitis being used) at Fast V 64 'nomen et aetatis mite [codd: rite Riese] senatus erat', 'the very name of senate signified a ripe old age' (Frazer).

134. PARENTIS = patris patriae. For the title compare Res Gestae 35 (the final achievement listed by Augustus) 'tertium decimum consulatum cum gerebam, senatus et equester ordo populusque Romanus uniuersus appellauit me patrem patriae, idque in uestibulo aedium mearum inscribendum esse et in curia et in foro Aug. sub quadrigis quae mihi ex s.c. positae sunt decreuit'. Suetonius describes the conferring of the title at Aug 58.

X. To Albinovanus Pedo

The poem combines with remarkable ease a number of quite disparate subjects, and is in this sense reminiscent of Tibullus. Most of the subjects had been used previously in the poetry of exile; in particular, see Tr I v 57-84 for an extended comparison of the trials of Ulysses and those of Ovid. The disquisition on the reasons for the Euxine's freezing over is, however, new. It seems to have been drawn from a geographical or physical treatise which has left its mark elsewhere in Latin literature: see at 37-38 (p 340-42).

1. CIMMERIO British Library Harley 2607 (Tarrant) CVMERIO M¹ IN ETIAM MEMORI C IN ********** B¹ IN HEMONIO HITP IN EVXINO F IN EXINO B^2c BISTONIO LM^2ul Many centuries had passed since the Cimmerians had inhabited Scythia; even Herodotus, who tells the story of their departure, seems to regard the event as belonging to the distant past (IV 11-12). Homer was vaguely aware of the nation: at Od XI 13-19 (imitated at Pan Mess 64-66), he speaks of the 'Κιμμερίων ἀνδρῶν ... πόλις' by the stream of Ocean, which never receives sunlight.

For Cimmerio Burman compared Claudian Cons Stil I 129 'nunc prope Cimmerii tendebat litora Ponti'; see as well In Eutr I 249 'extra Cimmerias, Taurorum claustra, paludes'.

1. BIS TERTIA ... AESTAS. The poem is therefore dated to the summer of 14. For Ovid's mentions of the length of his exile, see at vi 5 quinquennis (p 227).

3. ECQVOS ... ECQVOD Laurentianus 36 2, saec xv ET QVOS ... ET QVOD BCMFHILT. The same corruption is found in certain manuscripts at Met III 442-45 (Narcissus speaking) '"ecquis, io siluae, crudelius" inquit "amauit? ... ecquem ... qui sic tabuerit longo meministis in aeuo?"' and commonly. Other instances of ecquis in emotionally heightened questions at Fast IV 488, Tr I vi 11, EP III i 3, and Her XXI 106.

3. SILICES ... FERRVM. See at viii 49 tabida consumit ferrum lapidemque uetustas (p 270).

4. ALBINOVANE. Albinovanus Pedo[21] and Ovid seem to have been close friends. Ovid mentions him again at xvi 6 'sidereusque Pedo', and he was the source of the famous anecdote in the elder Seneca (Cont II 2 12) of how Ovid chose as the three lines in his poems he most wished to retain the same three verses a group of his friends most wished to remove.

He was a famous raconteur: the younger Seneca calls Pedo fabulator elegantissimus at Ep CXXII 15-16 when repeating one of his anecdotes.

At the time this poem was written, Albinovanus was engaged on a Theseid (71). Quintilian perhaps had this poem in mind when he included a rather slighting mention of Albinovanus in his catalogue of epic poets at X i 90: 'Rabirius ac Pedo non indigni cognitione, si uacet'. He may, however, have been thinking of Albinovanus' poem on Germanicus' campaigns, of which the elder Seneca preserves some twenty-three hexameters (Suas I 15; commentary by V. Bongi, Istituto Lombardo di scienze e lett. Rendiconti [Classe di Lettere] ser. 3 13 [1949], 28-48. Norden and others have attributed Morel Incert 46 'ingenia immansueta suoque simillima caelo' to the same poem). Martial several times mentions Albinovanus as a writer of epigrams (II lxxvii 5, V v 5 & X xx (xix) 10); this fits well with the younger Seneca's description of Albinovanus as fabulator elegantissimus.

At Ann I 60 2, Tacitus mentions Pedo as 'praefectus finibus Frisiorum' in Germanicus' campaign of 15.

5-6. LAPIDEM ... ANVLVS ... VOMER. See at viii 49 tabida consumit ferrum lapidemque uetustas (p 270), and compare AA I 473-76 'ferreus assiduo consumitur anulus usu, / interit assidua uomer aduncus humo. / quid magis est saxo durum, quid mollius unda? / dura tamen molli saxa cauantur aqua'.

6. ATTERITVR Heinsius. Korn and Riese printed the manuscripts' ET TERITVR, for which Riese cited Tr I iv 9-10 'pinea texta sonant pulsu [Rothmaler: pulsi codd], stridore rudentes, / ingemit et nostris ipsa carina malis' and Tr III iv 57-58 'ante oculos errant domus, urbsque et forma locorum, / acceduntque suis singula facta locis', but these are extended descriptions of single events, not lists of separate examples.

Elsewhere in Ovid, the only form found of atterere is attritus: this circumstance perhaps contributed to the corruption of the present passage.

6. ATTERITVR PRESSA VOMER ADVNCVS HVMO. Professor R. J. Tarrant points out to me the hypallage in this passage. Pressus is to be taken twice, with uomer and with humo: the earth is pressed down as the plough is pressed into it.

7. TEMPVS EDAX. The same phrase at Met XV 234; compare as well edax ... uetustas at Met XV 872.

7. PRAETER NOS. At EP II vii 39-45, Ovid (with a series of images parallel to that of the present passage) says that he is in fact being worn away by the hardships he is enduring: 'ut ... caducis / percussu crebro saxa cauantur aquis, / sic ego continuo Fortunae uulneror ictu ... nec magis assiduo uomer tenuatur ab usu, / nec magis est curuis Appia trita rotis, / pectora quam mea sunt serie calcata malorum'.

8. PERDIT I PERDET BCMFHLT. The tense is made probable by the preceding cauat ... consumitur ... atteritur and the following cessat; compare as well Tr IV vi 17-18 'cuncta potest ... uetustas / praeter quam curas attenuare meas'. Third conjugation verbs in the third person are for obvious reasons peculiarly apt to corruption of tense and mood. The alteration from present to future is rather less common than the inverse corruption, for an instance of which see at xii 18 reddet (p 378).

8. CESSAT DVRITIA MORS QVOQVE VICTA MEA. Death does not conquer Ovid, but is conquered by him. Professor E. Fantham points out to me the baroque inversion in the phrase, citing as a parallel Sen Tr 1171-75, where Hecuba says that death fears her and flees her.

Riese placed a question mark at the end of the line, but since in 7 Ovid asserts unambiguously that time does not affect him, there seems no reason to make the following line a question. In his poems from exile Ovid often expresses his wish to die; see Tr III viii 39-40 'tantus amor necis est querar ut cum Caesaris ira / quod non offensas uindicet ense suas', Tr III xiii 5-6, IV vi 49-50, and V ix 37-38.

9. EXEMPLVM EST ANIMI NIMIVM PATIENTIS VLIXES. Ovid frequently compares his trials in exile to those undergone by Ulysses. The longest instance of this is Tr I v 57-84; compare as well Tr III xi 61-62 'crede mihi, si sit nobis collatus Vlixes, / Neptuni minor est quam Iouis ira fuit', Tr V v 1-4, and EP I iii 33-34, II vii 59-60 & III vi 19-20.

Ulysses' voyage was a favourite subject of the Latin poets. For a surviving example, see Prop III xii 23-36. An indication of the subject's popularity is the fact that Pan Mess 45-49 'nam seu diuersi fremat inconstantia uulgi, / non alius sedare queat; seu iudicis ira / sit placanda, tuis poterit mitescere uerbis. / non Pylos aut Ithace tantos genuisse feruntur / Nestora uel paruae magnum decus urbis Vlixem' is followed not by a description of Ulysses' eloquence, as would have been appropriate, but by a narrative of his travels (52-81): this illogical sequence was no doubt induced by the poet's familiarity with similar descriptions of Ulysses' voyage in the poetry of his time.

Professor E. Fantham cites Seneca's use of Ulysses as an exemplum patientiae at Sen Dial II 2 1, where Hercules is compared to Ulysses.

9. EXEMPLVM EST. Professor R. J. Tarrant points out to me the unusual baldness of the phrase. In Ovid's earlier verse exemplum has an instructional or minatory overtone (AA III 686, Met IX 454). The flatter use of exemplum seems to be typical of the poetry of exile: compare EP III i 44 'coniugis exemplum diceris esse bonae', and Tr I v 21, IV iii 72 & IV iv 71.

9. NIMIVM PATIENTIS = πολύτλας (Il VIII 97, Od V 171, et saep.). The sense of nimium seen here is not generally found in poetry, or even in literary prose; the instances cited by OLD nimium² 2 are all from comedy, Cato, and the letters of Cicero.

10. DVO LVSTRA. Compare xvi 13-14 'Vlixem / errantem saeuo per duo lustra mari' and AA III 15-16 'est pia Penelope lustris errante duobus / et totidem lustris bella gerente uiro'.

11. SOLLICITI ... FATI is based on such phrases as sollicita uita (Prop II vii 1) and sollicitissima aetas (Sen Breu Vit 16 1). Similar phrasing at Tr IV x 116 'nec me sollicitae taedia lucis habent'.

11. PLACIDAE SAEPE FVERE MORAE. Compare Prop III xii 23-24 'Postumus alter erit miranda coniuge Vlixes: / non illi longae tot nocuere morae'.

13. SEX ANNIS. According to Homer (Od VII 261), Ulysses left Calypso in the eighth year of his stay on her island. André points out that Hyginus Fab CXXV 16 has Ulysses on the island for one year only; for other estimates of the length of Ulysses' stay, see Roscher III 627. Ovid was probably influenced by the bis ... tertia of the poem's opening. Cimmerio in 1 furnishes another connection with Ulysses (Od XI 14; quoted at 1).

13. FOVISSE. Compare Od V 118-120 (Calypso speaking) 'Σχέτλιοί ἐστε, θεοί, ζηλήμονες ἔξοχον ἄλλων, / οἵ τε θεαῖς ἀγάασθε παρ' ἀνδράσιν εὐνάζεσθαι / ἀμφαδίην, ἤν τίς τε φίλον ποισετ' ἀκοίτην'.

13. CALYPSO BCMILT. Lenz and André print CALYPSON (FH). Roman poets followed the Greek declension of feminine proper nouns ending in -ω; compare Pan Mess 77 'fecunda Atlantidos arua Calypsus [uar calipsos]'. The accusatives of such nouns are of the same form as the nominative. See for example Aen IV 383-84 'et nomine Dido / saepe uocaturum' and Aen VII 324-25 'luctificam Allecto dirarum ab sede dearum / infernisque ciet tenebris', cited by Charisius 63 (Keil); neither he nor Servius shows knowledge of an accusative in -on. Scribes, however, found the declension puzzling; and it is common to find the pseudo-accusative in -on offered by some manuscripts whenever the true form in -o occurs; this has happened at Her VI 65 'ultimus e sociis sacram conscendis in Argo', Her VII 7 'certus es ire tamen miseramque relinquere Dido [edd: Didon codd]', Her XII 9 'cur umquam Colchi Magnetida uidimus Argo', Am II ii 45 'dum nimium seruat custos Iunonius Io', Am II xix 29 'dum seruat Iuno mutatam cornibus Io', and Prop I xx 17-18 'namque ferunt olim Pagasae naualibus Argo [edd: Argon codd] / egressam longe Phasidos isse uiam'. Modern editors often print the spurious form, even at AA I 323 'et modo se Europen fieri, modo postulat Io', where all manuscripts offer the correct reading.

For a full discussion of this and the inverse corruption (for instance of Iason to Iaso), see Goold 12-14.

14. AEQVOREAEQVE. Compare Am II xvii 17-18 'creditur aequoream Pthio Nereida regi, / Egeriam iusto concubuisse Numae' and AA II 123-24 'non formosus erat, sed erat facundus Vlixes, / et tamen aequoreas torsit amore deas'. Merkel's AEAEAEQVE is ingenious but unnecessary.

15. HIPPOTADES = Aeolus. The same patronymic at Met IV 663, XI 431, XIV 86, XIV 224 & XV 707.

15. QVI DAT PRO MVNERE VENTOS. Compare Met XIV 223-26 'Aeolon ille refert Tusco regnare profundo, / Aeolon Hippotaden, cohibentem carcere uentos; / quos bouis inclusos tergo, memorabile munus, / Dulichium sumpsisse ducem' and Od X 19-26.

17. NEC BENE CANTANTES LABOR EST AVDISSE PVELLAS. The description is intentionally prosaic. For the Homeric account of the Sirens see Od XII 37-54 & 153-200.

17. AVDISSE F AVDIRE BCMHILT. Audire cannot stand, as the present tense conflicts with fuit in the following line. For est audisse representing fuit audire, compare Met IX 5-6 (Achelous hesitates before recounting his wrestling-match with Hercules) 'referam tamen ordine: nec tam / turpe fuit uinci quam contendisse decorum est'.

18. NEC DEGVSTANTI LOTOS AMARA FVIT. See Od IX 82-104 for Homer's account of the Lotus-eaters.

18. NEC ... AMARA = et dulcis. Compare Od IX 94 'λωτοῖο ... μελιηδέα καρπόν'.

18. DEGVSTANTI. The verb is extremely rare in the sense 'taste, sample'; this is the only instance of the meaning found in poetry, although a transferred use is found at Lucretius II 191-92 'ignes ... celeri flamma degustant tigna trabesque' and Aen XII 375-76 'lancea ... summum degustat uulnere corpus'.

Ovid uses the somewhat more common gustare in a similar context at Tr IV i 31-32 'sic noua Dulichio lotos gustata palato / illo quo nocuit grata sapore fuit'.

21. VRBEM LAESTRYGONOS = 'Λάμου αἰπὺ πτολίεθρον, / Τηλέπυλον Λαιστρυγονίην' (Od X 81-82) or 'Lami ueterem Laestrygonos ... urbem' (Met XIV 233), where the crews of all the ships but Ulysses' own were killed and eaten; accounts of this at Od X 76-132 and Met XIV 233-42. Ovid refers again to the episode at EP II ix 41 'quis non Antiphaten Laestrygona deuouet?'.

21. LAESTRYGONOS BC LE(-I-)STRYGONIS MFHILT. Laestrygonos = Λαιστρυγόνος (Od X 106). At Met XIV 233 (cited above) all manuscripts offer Laestrygonis; the Greek genitive should probably be read as here.

22. GENTIBVS OBLIQVA QVAS OBIT HISTER AQVA. Similar wording at ii 37-38 'hic mea cui recitem nisi flauis scripta Corallis, / quasque alias gentes barbarus Hister obit?'.

22. OBLIQVA apparently refers to the swirling of a river's eddies. The sense 'winding' generally given the word would fit at Met IX 17-18 (Achelous to the father of Deianira) 'dominum me cernis aquarum / cursibus obliquis inter tua regna fluentum', but not at Met VIII 550-53 (Achelous to Theseus) '"succede meis" ait "Inclite, tectis, / Cecropide, nec te committe rapacibus undis: / ferre trabes solidas obliquaque uoluere magno / murmure saxa solent"' or Her VI 87 'illa refrenat aquas obliquaque flumina sistit'. At Met I 39 'fluminaque obliquis cinxit decliuia ripis', obliquis should be taken with flumina, and decliuia with ripis; or possibly both adjectives should be taken with both nouns.

23. VINCET. Like superare, uincere has the twin meanings of 'surpass' and 'defeat'.

23. CYCLOPS. The same pairing of the Laestrygonians and Polyphemus at EP II ii 113-114 (to Messalinus; he should address Augustus on Ovid's behalf) 'nec tamen Aetnaeus uasto Polyphemus in antro / accipiet uoces Antiphatesue tuas'.

23. FERITATE goes with uincet: 'will surpass in savagery'. I once thought PIETATE (BCI^ac) was the correct reading, connecting the word with saeuum and taking it as a reference to human sacrifice; but this seems strained and obscure. Pietate may be an intrusion from ecclesiastical Latin; Professor R. J. Tarrant suggests that it is possibly an anticipation of the following Piacchen.

23. PIACCHEN B PIAECHEN C. See the critical apparatus for the other forms offered by the manuscripts. As the king's name is not elsewhere recorded, its true form must remain in doubt.

24. QVI QVOTA TERRORIS PARS SOLET ESSE MEI. With Burman, Weber, and Wheeler I take the line as a statement: compare EP II x 31 'et quota pars haec sunt rerum quas uidimus ambo' (cited by Williams), where quota, as here, takes the meaning 'how small' from context. Most editors take it as a question, for which compare Am II xii 9-10 'Pergama cum caderent bello superata bilustri, / ex tot in Atridis pars quota laudis erat?'.

25-27. SCYLLA ... CHARYBDIN. Ovid gives similar descriptions of Scylla at Am III xii 21-22 and EP III i 122, of Charybdis at Am II xvi 25-26, and of Scylla and Charybdis at Her XII 123-26 and Met XIII 730-33. All such descriptions in Latin poetry of course derive ultimately from Od XII 73-110.

25. QVOD LATRET AB INGVINE MONSTRIS. Professor R. J. Tarrant points out to me Ovid's imitation here of Ecl VI 74-75 'Scyllam ... candida succinctam latrantibus inguina monstris'; the rates and nautae of Ovid's line 26 are in lines 76 and 77 of the Virgilian passage.

25. QVOD. 'Granted that'. Bömer at Met VII 705 claims that the only passage where this is the necessary meaning of quod is Priapea VI 1 'quod sum ligneus ... Priapus ... prendam te tamen', but it seems to be the meaning required at Lucretius II 532-35 'nam quod rara uides magis esse animalia quaedam / fecundamque minus naturam cernis in illis, / at regione locoque alio terrisque remotis / multa licet genere esse in eo numerumque repleri'.

All six instances of the idiom cited by the OLD (quod 6c) are from poetry. In the two instances already cited, quod is followed by the indicative, as is the case at Prop III ii 11-16. Quod in this sense followed by the subjunctive seems to be an Ovidian idiom; it is used by him at Her IV 157-61 'quod mihi sit genitor, qui possidet aequora, Minos, / quod ueniant proaui fulmina torta manu, / quod sit auus radiis frontem uallatus acutis, / purpureo tepidum qui mouet axe diem— / nobilitas sub amore iacet!' and Met VII 704-7 'liceat mihi uera referre / pace deae: quod sit roseo spectabilis ore, / quod teneat lucis, teneat confinia noctis, / nectareis quod alatur aquis, ego Procrin amabam', and by an imitator of Ovid at Her XVIII 41.

26. HENIOCHAE NAVTIS PLVS NOCVERE RATES. The Heniochi lived on the eastern shore of the Euxine and were, as Ovid indicates, known as pirates (Strabo XI 2 12-13).

27. INFESTIS ... ACHAEIS. Mela includes the Achaei and the Heniochi in his list of 'ferae incultaeque gentes uasto mari adsidentes' (I 110). The two nations are grouped together by Strabo (XII 2 12) and Pliny (NH VI 30).

28. EPOTVM ... VOMAT. Professor R. J. Tarrant cites the verbal similarity at (pseudo-Ovidian) Am III v 18 'iterum pasto pascitur ante cibo'.

28. EPOTVM B ET POTVM C EPOTET MFHILT. Epotet is supported by Her XII 125 'quaeque uomit totidem fluctus totidemque resorbet' and Od XII 105-6 'τρὶς μὲν γάρ τ' ἀνίησιν ἐπ' ἤματι, τρὶς δ' ἀναροιβδεῖ / δεινόν'. Professor A. Dalzell points out in particular 'τρὶς ... τρὶς' paralleling ter ... ter in the present passage. But at RA 740 Ovid wrote 'hic uomit epotas [uarr et potat; hic potat; optatas; acceptas; aequoreas] dira Charybdis aquas'; and the corruption to epotet seems much more probable than the inverse. Ovid elsewhere uses only the perfect participle of epotare.

29. LICENTIVS ERRANT. Ovid is clearly imitating Aen VII 557-58 (Juno to Allecto) 'te super aetherias errare licentius auras / haud pater ille uelit, summi regnator Olympi', apparently the only other instance of licentius in classical verse.

31-32 act as a bridge to the next major section of the poem, and do not in themselves contribute to what has been said.

31. INFRONDES is a hapax legomenon.

32. HIC FRETA VEL PEDITI PERVIA REDDIT HIEMPS. Other mentions of the sea's freezing at vii 7, Tr II 196, III x 35-50 & V x 2, and EP III i 15-16 (to the Pontus) 'tu glacie freta uincta tenes, et in aequore piscis / inclusus tecta saepe natauit aqua'.

Parts of the Black Sea do in fact freeze: 'In winter, spurs of the Siberian anticyclone (clear, dry, high-pressure air mass) create a strong current of cold air, and the northwestern Black Sea cools down considerably, with regular ice formation' (article on "Black Sea", Encyclopaedia Britannica, Macropaedia vol. 2, pp. 1096-98 [Chicago: 1974]).

32. HIEMPS. For the last one hundred years, the spelling given in editions of Latin texts has generally been hiems (some exceptions are Palmer's Heroides, the Paravia Virgil, and Reynolds' editions of Seneca), but the spelling in the ancient manuscripts of Virgil is invariably hiemps. Munro's argument for this spelling seems unanswerable: 'obeying the almost unanimous testimony of our own [i.e. O and Q of Lucretius] and other good mss. we cannot but give umerus umor and the like: also hiemps. I have heard it asked what then is the genitive of hiemps; to which the best reply perhaps would be what is the perfect of sumo or the supine of emo. The Latins wrote hiemps, as they wrote emptum sumpsi sumptum and a hundred such forms, because they disliked m and s or t to come together without the intervention of a p sound; and our mss. all attest this: tempto likewise is the only true form, which the Italians in the 15th century rejected for tento' (Lucretius ed. 4 vol. 1 p. 33).

33-34. VT, QVA REMVS ITER PVLSIS MODO FECERAT VNDIS, / SICCVS CONTEMPTA NAVE VIATOR EAT. Ovid has in mind Virgil's description of the freezing of a Scythian river (G III 360-62) 'concrescunt subitae currenti in flumine crustae, / undaque iam tergo ferratos sustinet orbis, / puppibus illa prius, patulis nunc hospita plaustris'.

35. QVI VENIVNT ISTINC VIX VOS EA CREDERE DICVNT; / QVAM MISER EST QVI FERT ASPERIORA FIDE. For Ovid's fear that his accounts of what he has undergone will not be believed, see vii 3-4 and Tr I v 49-50, III x 35-36 & IV i 65-66. In particular, see ix 85-86 'mentiar, an coeat duratus frigore Pontus, / et teneat glacies iugera multa freti'.

37-38. NEC TE CAVSAS NESCIRE SINEMVS / HORRIDA SARMATICVM CVR MARE DVRET HIEMPS. Ovid's principal explanation of the freezing of the Euxine, the low salinity of the water, is found in four other Latin authors. At IV 718-28, Valerius Flaccus offers a catalogue of rivers similar to that of Ovid, and, like Ovid, gives the cold winter winds as a subsidiary reason for the freezing. It is quite possible that Ovid is Valerius' source; but this is very unlikely to be the case for Macrobius Sat VII xii 28-38 (cited by Burman). The passage is a discussion of why, although oil congeals, wine and vinegar do not. Wine does not freeze because it contains elements of fire; this is why Homer called it αἴθοπα οἶνον. Vinegar does not freeze because it is so bitter; it is like seawater, which because of its bitterness does not congeal. 'nam quod Herodotus historiarum scriptor contra omnium ferme qui haec quaesiuerunt opinionem scripsit [IV 28], mare Bosporicum, quod et Cimmerium appellat, earumque partium mare omne, quod Scythicum dicitur, id gelu constringi et consistere, aliter est quam putatur'. It is not the seawater that freezes, but the layer of fresh water above it, which comes from the rivers that flow into the Euxine. Macrobius goes on to explain that there is an outflow of fresh water to the Mediterranean and an influx of seawater, with perfect correctness: the Encyclopaedia Britannica article cited at 32 notes that 'Flows in the Bosporus are complex, with surface Black Sea water going out and deep, saltier water coming in from the Sea of Marmara*.

There can be very little doubt, given the identity of the explanations and the similarity of language, that Ovid and Macrobius were drawing on a common source. The same source is reflected at Gellius XVII viii 8-16. Here Taurus the philosopher asks Gellius why oil often congeals, but wine does not. Gellius answers that wine is fiery by nature, which is why Homer called it αἴθοπα οἶνον. Taurus responds that wine is indeed known to have fire in it, for it warms the body when drunk; yet vinegar, in spite of its cooling effects, never freezes; perhaps things which are light and smooth are more prone to freezing. It is also worth asking why fresh water freezes, but seawater does not. 'tametsi Herodotus ... historiae scriptor contra omnium ferme qui haec quaesiuerunt opinionem scribit mare Bosporicum, quod Cimmerium appellatur, earumque partium mare omne quod Scythicum dicitur, gelu stringi et consistere'. No explanation for the freezing-over is given.[22]

Ammianus Marcellinus XXII 8 48 gives the same two explanations for the Euxine's freezing as Ovid: 'quicquid autem eiusdem Pontici sinus Aquilone caeditur et pruinis, ita perstringitur gelu ut nec amnium cursus subteruolui credantur, nec per infidum et labile solum gressus hominis possit uel iumenti firmari, quod uitium numquam mare sincerum, sed permixtum aquis amnicis temptat'. At XXII 8 46 he once again mentions the sweetness of the Euxine's waters.

Lucan describes the freezing of the Euxine (V 436-41), but gives no explanation of the cause.

39. PLAVSTRI PRAEBENTIA FORMAM ... SIDERA. The Great Bear. Other mentions of the constellation at Met X 446-47 'inter ... triones / flexerat obliquo plaustrum temone Bootes', Tr III iv b 1-2 (47-48), III x 3-4 & V iii 7-8, and EP I v 73-74. Compare as well Germanicus Aratea 24-26 'axem Cretaeae dextra laeuaque tuentur / siue Arctoe seu Romani cognominis Vrsae / Plaustraue [Grotius:-que codd], quae facie [scripsi (datiuum)[23]: facies codd] stellarum proxima uerae [Barth: uera uel uero codd]', Her XVIII 152, Sen Ag 66-68, and Lucan V 23 'Hyperboreae plaustrum glaciale sub Vrsae'.

Praebentia formam is elevated diction: Professor R. J. Tarrant cites Lucretius V 581-83 'luna ... claram speciem certamque figuram / praebet'.

40. PERPETVVM M^2ul PRAECIPVVM BCM¹FHILT. Praecipuum could be defended by EP III i 13-14 (to the Pontus) 'nec tibi pampineas autumnus porrigit uuas, / cuncta sed immodicum tempora frigus habet', but praecipuus in fact always seems to have the notion of 'outstanding' or 'superior', which does not seem appropriate to the present passage. For perpetuum compare Tr III ii 7-8 'plurima sed pelago terraque pericula passum / ustus ab assiduo frigore Pontus habet', Tr III x 14 '[niuem ...] indurat Boreas perpetuamque facit', Tr V ii 65-66 'me ... cruciat numquam sine frigore caelum, / glaebaque canenti semper obusta gelu', EP I iii 49-50 'orbis in extremi iaceo desertus harenis, / fert ubi perpetuas obruta terra niues', and EP II vii 72 'frigore perpetuo Sarmatis ora riget'.

41. HINC ORITVR BOREAS. Compare Tr III xi 7-8 'barbara me tellus et inhospita litora Ponti / cumque suo Borea Maenalis ursa uidet' and Ibis 11-12 'ille relegatum gelidos Aquilonis ad ortus / non sinit exilio delituisse meo'.

41. DOMESTICVS. The word is rare in verse; Ovid uses it as a substantive at iii 15 'ille ego conuictor densoque domesticus usu'. Here Ovid may be recalling the language of Met VI 685-86 (of Boreas) 'ira, / quae solita est illi nimiumque domestica uento'.

42. VIRES. Merkel proposed MORES, citing Virgil G I 50-52 'at prius ignotum ferro quam scindimus aequor, / uentos et uarium caeli praediscere morem / cura sit' and Statius Sil III ii 87 'quos tibi currenti praeceps gerat Hadria mores'. The second passage is not to the point, since it means 'what sort of obedience to your wishes do you expect from the Adriatic as you make your voyage'. In any case, Professor R. J. Tarrant points out to me the poor logic of Merkel's proposed text: Ovid is deriving the natura loci from its surroundings; he should not now be saying that Boreas gets his mores from the area. The reading of the manuscripts seems acceptable enough if one accepts Meynke's polo for loco ('he gathers strength from the nearby North Pole'). For sumit uires compare Met VIII 882 (Achelous speaking) 'armenti modo dux uires in cornua sumo', Met XI 510-11 'ut ... solent sumptis incursu uiribus ire ... feri ... leones' and Hor Ep I xviii 85 'neglecta solent incendia sumere uires'. Professor R. J. Tarrant compares such phrases as sumere iras (Met II 175), animos (Met III 544-45), and cornua (AA I 239, Tr IV ix 27).

42. POLO Meynke LOCO codd. The pointlessness of loco is made clear enough by Wheeler's 'and he takes on strength from a place nearer to him'. Meynke's polo removes the difficulty, answers well to the following 'at Notus, aduerso tepidum qui spirat ab axe', and is supported by the language of Met II 173 'quaeque polo posita est glaciali proxima Serpens', and Fast IV 575-76 (of Ceres) 'errat et in caelo liquidique immunia ponti / adloquitur gelido proxima signa polo'. For the corruption, compare the common misreading of locum for solum.

43. ADVERSO ... AB AXE. Ovid here seeks a contrast with polo in the previous line; but clearly he means only that the south wind comes from the opposite direction, not that it originates at the South Pole.

Bentley conjectured AVERSO for aduerso, and the two words are obviously prone to interchange: compare Tr I iii 45 (of Ovid's wife, after his departure) 'multaque in auersos [Heinsius: aduersos codd] effudit uerba Penates' and the variations among the manuscripts at Virgil G I 218 'auerso ... astro', Aen XII 647 'auersa uoluntas', and Sen Tr 1123 'auersa cingit campus' (on which see Housman 1076). But aduerso 'opposite' seems to have the sense required here.

43. TEPIDVM QVI SPIRAT. For the construction compare Met IX 661 'sub aduentu spirantis lene Fauoni' and Avienus Descr Orb 847 'uel qua lene Notus spirat'. The trivialized TEPIDVS QVI SPIRAT is found in MH^2c. Tepidus Notus occurs four times in Ovid (Am I iv 12, I vii 56 & II viii 20, and Tr III xii [xiii] 42).

44. LANGVIDIORQVE VENIT. Compare EP II i 1-2 'Huc quoque Caesarei peruenit fama triumphi, / languida quo fessi uix uenit aura Noti'.

46. AB AMNE. Similar instrumental uses of ab at Her X 138 'tunicas lacrimis sicut ab imbre graues', AA III 545 'ingenium placida mollitur ab arte', Met I 65-66 'contraria tellus / nubibus assiduis pluuiaque madescit ab Austro', Met IV 162-63 'pectus ... adhuc a caede tepebat', and Fast V 323 'caelum nigrescit ab Austris'.

47-58. For the lengthy catalogue, typical of Ovid, compare the listing of Actaeon's dogs at Met III 206-25 (in particular at 217 'et Dromas et Canache Sticteque et Tigris et Alce') and the catalogue of trees that came to listen to Orpheus sing (Met X 90-107).

47. LYCVS. A number of rivers had this name in the ancient world. Ovid presumably means the Paphlagonian Lycus referred to by Virgil at G IV 366-67 'omnia sub magna labentia flumina terra / spectabat diuersa locis, Phasimque Lycumque ...'.

47. SAGARIS. The modern Sakarya; it flows into the Black Sea about 125 kilometres east of Istanbul. It is mentioned at Pliny NH VI 1 4 'Sangaris fluuius ex inclutis. oritur in Phrygia, accipit uastos amnes ... idem Sagiarius plerisque dictus'.

47. PENIVSQVE. The 'flumen et oppidum Penius' are mentioned at Pliny NH VI 14 as being in the region of the Caucasus on the Euxine coast; nearby were 'multis nominibus Heniochorum gentes'. The river seems not to be mentioned elsewhere in ancient literature.

47. HYPANISQVE. The modern Bug empties into the Black Sea about 50 kilometres east of Odessa. It is mentioned again by Ovid at Met XV 285-86 'quid? non et Scythicis Hypanis de montibus ortus, / qui fuerat dulcis, salibus uitiatur amaris?' and Virgil G IV 370 'saxosumque sonans Hypanis'.

47. CALESQVE. Isaac Vossius made this correction for the manuscripts' CATESQVE (I has CHARESQVE) on the basis of 'Eustathio Scholiis in Periegeten'. Heinsius aptly cited a description of the occasionally violent flow of the river at Thucydides IV 75 2.

As indicated by this passage, the modern Alapli flows into the Black Sea near Ereğli, about 200 kilometres east of Istanbul.

48. CREBRO VERTICE TORTVS HALYS. An imitation of Aen VII 566-67 'fragosus / dat sonitum saxis et torto uertice torrens'. Tortus when used of water generally refers to the disturbance caused by rowing (Fast V 644; Catullus LXIV 13; Aen III 208).

48. HALYS. The modern Kizil Irmak flows into the Black Sea about 600 kilometres east of Istanbul. André compares Apollonius' description of the river (II 366-67) 'ῥοαὶ Ἅλυος ποταμοῖο / δεινὸν ἐρεύγονται'.

49-50. The three rivers mentioned in these lines are all named for their swiftness.

49. PARTHENIVSQVE RAPAX. The modern Bartin flows into the Black Sea about 280 kilometres east of Istanbul and about 240 kilometres west of Sinop. It is in fact a very calm river: this information was available to Ovid from Apollonius II 936-37 'Παρθενίοιο ῥοὰς ἁλιμυρήεντος, / πρηυτάτου ποταμοῦ' (cited by André).

49. VOLVENS SAXA. Similar phrasing at Met VIII 552-53 '[undae ...] ferre trabes solidas obliquaque uoluere magno / murmure saxa solent'.

49. CINAPSES BC CINAPSIS L TYNAPSES H CINASPES FIT NIPHATES M. Editors read CYNAPSES; but since the river is not otherwise known, restoration is dangerous. M's reading looks like an interpolation from Lucan III 245 'Armeniusque tenens uoluentem saxa Niphaten' (cited by Micyllus).

50. NVLLO TARDIOR = uelocior omni; André mistranslates 'le plus lent des fleuves'. Compare Tr I v 1 'O mihi post nullos umquam [uar ullos numquam] memorande sodales' and EP I iii 65-66 'Zmyrna uirum tenuit, non Pontus et hostica tellus, / paene minus nullo Zmyrna petenda loco'.

50. TYRAS. The modern Dnestr flows into the Black Sea about fifty miles south of Odessa; near its mouth is the city of Ovidiopol. The river is briefly mentioned at Pliny NH IV 82 & 93, and at Mela II 7, where it is called the 'Tyra'; this however seems to be a scribal error induced by the following separat.

51. THERMODON. The modern Terme flows into the Black Sea about 100 kilometres southeast of the mouth of the Kizil Irmak (Halys). It was conventional to mention the Amazons in connection with the river (Met XII 611, Aen XI 659-60, Prop III xiv 13-14, Ammianus Marcellinus XXII 8 17). Professor E. Fantham suggests to me that Ovid may here be providing Albinovanus with material for the part of his Theseid dealing with Theseus' expedition against the Amazons.

Ovid also mentions the Thermodon at Met I 248-49 (the story of Phaethon) 'arsit et Euphrates Babylonius, arsit Orontes / Thermodonque citus Gangesque et Phasis et Hister'. As in the present distich, the Thermodon and Phasis, both prominent in mythology, are mentioned together.

51. TVRMAE BCM TVRBAE FHILT. There is a similar variation among the manuscripts at AA III l-2 'Arma dedi Danais in Amazonas; arma supersunt / quae tibi dem et turmae, Penthesilea, tuae'. From other descriptions of the Amazons, the Auctor Electorum Etonensium aptly compares Val Fl IV 603 (cateruas) and 607 (turma); compare as well Statius Sil I vi 56 (turmas). It is possible that turma should be read at Prop III xiv 13-14 'qualis Amazonidum nudatis bellica mammis / Thermodontiacis turba lauatur aquis'; but this would make bellica redundant.

53. BORYSTHENIO ... AMNE = Bory̅sthĕnē. The river is the modern Dnepr, which flows into the Black Sea about 120 kilometres east of Odessa, about 50 kilometres east of the mouth of the Bug (Hypanis). For the metrical device here employed, compare Prop II vii 17-18 'hinc etenim tantum meruit mea gloria nomen, / gloria ad hibernos lata Borysthenidas', Avienus Descr Orb 448 'inde Borysthenii uis sese fluminis effert' & 721 'ora Borysthenii qua fluminis in mare uergunt'.

53. LIQVIDISSIMVS is not found elsewhere in Ovid.

53. DIRAPSES. The river is not mentioned elsewhere.

54. MELANTHVS. The modern Melet Irmak flows into the Black Sea about 25 kilometres west of Trabzon (Trapezus). It is mentioned in passing at Pliny NH VI 11.

55-56. QVIQVE DVAS TERRAS, ASIAM CADMIQVE SOROREM, / SEPARAT ET CVRSVS INTER VTRAMQVE FACIT. The Tanais (Don) is named as the border between Europe and Asia by Pliny (NH IV 78) and Avienus (Descr Orb 28 & 861). Compare as well Lucan III 272-76 'qua uertice lapsus / Riphaeo Tanais diuersi nomina mundi / imposuit ripis Asiaeque et terminus idem / Europae, mediae dirimens confinia terrae, / nunc hunc, nunc illum, qua flectitur, ampliat orbem'.

Vibius Sequester (Geog Lat min [Riese] p. 212) has an entry 'Hypanis Scythiae qui, ut ait Gallus "uno tellures diuidit amne duas": Asiam enim ab Europa separat'. The Hypanis cannot be the river Ovid is here referring to, for it has already been mentioned in 47; but, as Lenz saw, the line from Gallus could well have been in Ovid's mind as he wrote this passage. Professor R. J. Tarrant notes that the extraordinary Cadmique sororem could well be a borrowing from the earlier poet.

57-58. INTER MAXIMVS OMNES / CEDERE DANVVIVS SE TIBI, NILE, NEGAT. A similar conjunction at Tr III x 27-28 'ipse, papyrifero qui non angustior amne, / miscetur uasto multa per ora freto'. Herodotus compares the courses of the Nile and the Danube, concluding 'οὕτω τὸν Νεῖλον δοκέω διὰ πάσης τῆς Λιβύης διεξιόντα ἐξισοῦσθαι τῷ Ἴστρῳ' (II 34), referring to the length of the rivers, however, rather than their volume of discharge. At NQ III 22 Seneca mentions the belief of some that because of their large size and the fact that their sources were both unknown the Nile and the Danube must both have been formed at the creation of the world, unlike other rivers. At IV 1 1-2 he argues against those who equated the two rivers, pointing out that the source of the Danube was known to be in Germany, and that the two rivers flood at different times of the year.

59. COPIA TOT LATICVM QVAS AVGET ADVLTERAT AQVAS. The comparative freshness of the waters of the Black Sea was well known in antiquity. Besides the passages cited at 37-38, see Polybius IV 42 3 and Philostratus Imag I 13 7.

61-62. QVIN ETIAM, STAGNO SIMILIS PIGRAEQVE PALVDI, / CAERVLEVS VIX EST DILVITVRQVE COLOR. Ovid's drinking water was, on the other hand, rather brackish: 'est in aqua dulci non inuidiosa uoluptas: / aequoreo bibitur cum sale mixta palus' (EP II vii 73-74).

63. INNATAT VNDA FRETO DVLCIS. Similar wording at Macrobius Sat VII 12 32 'superficies maris, cui dulces aquae innatant, congelascit'.

64. PONDVS B¹CMFHT NOMEN ILB². Wakefield conjectured MOMEN on the basis of Lucretius VI 473-74 'quo magis ad nubis augendas multa uidentur / posse quoque e salso consurgere momine ponti'. But pondus seems appropriate to the context in a way that momen 'heaving' does not. Nomen habe(n)t is a frequent line-ending in Ovid, occurring some twenty-five times (once in Her XVI). Proprium nomen occurs in Ovid at Fast V 191-92 (Ovid is addressing Flora) 'ipsa doce quae sis. hominum sententia fallax: / optima tu proprii nominis auctor eris' and EP I viii 13-14 'Caspius Aegissos, de se si credimus ipsis, / condidit et proprio nomine dixit opus'. The phrase would have been very familiar to the scribes from grammatical treatises ('proper noun'). A combination of these circumstances no doubt induced the error.

Professor A. Dalzell suggests to me that momen is perhaps correct, the notion being that the salt water keeps moving, and so does not freeze. Pondus would then be a (mistaken) gloss that has displaced momen from the text; nomen would be a simple misreading of momen.

66. CERTIS ... MODIS. 'Metre'; compare Fast III 388 'ad certos uerba canenda modos', Tib II i 51-52 'agricola ... primum ... cantauit certo rustica uerba pede' and Manilius III 35 'pedibus ... iungere certis'.

67. DETINVI ... TEMPVS, CVRASQVE FEFELLI excerpta Politiani DETINVI ... TEMPVS CVRAMQVE FEFELLI LT DETINVI ... CVRAS TEMPVSQVE FEFELLI BCMFHI. Tempus fallere 'make time pass unnoticed' is perfectly acceptable Latin; compare Tr III iii 11-12 'non qui labentia tarde / tempora narrando fallat amicus adest', Her I 9-10 'nec mihi quaerenti spatiosam fallere noctem / lassaret uiduas pendula tela manus', Met VIII 651 'interea medias fallunt sermonibus horas', Tr IV x 112-14 'tristia ... carmine fata leuo. / quod quamuis nemo est cuius referatur ad aures, / sic tamen absumo decipioque diem', and Her XIX 37-38 'tortaque uersato ducentes stamina fuso / feminea tardas fallimus arte moras'. The difficulty with the manuscript reading in the present passage is that detinui curas is without parallel. Heinsius therefore accepted Politian's reading, citing in its support Met I 682-83 'sedit Atlantiades et euntem multa loquendo / detinuit sermone diem'. The Auctor Electorum Etonensium objected that detinui tempus was inappropriate: 'poeta tempus detinere noluit, quod scilicet per se morari atque haerere uidebatur inuisum'. He conjectured DISTINVI CVRAS and Burman DIMINVI CVRAS, which he later found in one of his manuscripts. But detinere here can have the same meaning 'occupy, keep busy' as it has at the Metamorphoses passage, where A. G. Lee cites the present passage (with Politian's reading) and Tr V vii 39 'detineo studiis animum falloque dolores'.

The interchange of adjoining metrically and grammatically equivalent substantives is very common.

67-68. "DETINVI" DICAM "TEMPVS, CVRASQVE FEFELLI; / HVNC FRVCTVM PRAESENS ATTVLIT HORA MIHI". The thought of the passage also at ii 39-40 & 45 'quid nisi Pierides, solacia frigida, restant', Tr V i 33-34 'tot mala pertulimus, quorum medicina quiesque / nulla nisi in studio est Pieridumque mora', and EP I v 53-55 'magis utile nil est / artibus his, quae nil utilitatis habent. / consequor ex illis casus obliuia nostri'.

69. ABFVIMVS SOLITO ... DOLORE. Compare Cic Fam IV iii 2 'a multis et magnis molestiis abes'; I have found no parallel from verse.

71. CVM THESEA CARMINE LAVDES. See at 4 Albinouane (p 327).

71. THESEA. For Theseus as the type of loyalty, compare Tr I iii 66 'o mihi Thesea pectora iuncta fide!', I v 19-20, I ix 31-32, V iv 25-26 (Ovid's letter speaking) 'teque Menoetiaden, te qui comitatus Oresten, / te uocat Aegiden Euryalumque suum', and EP II iii 43, II vi 26 & III ii 33-34 'occidit et Theseus et qui comitauit Oresten; / sed tamen in laudes uiuit uterque suas'. From other authors, Otto Theseus cites Prop II i 37-38, Martial VII xxiv 3-4 & X xi 1-2, Claudian Ruf I 107, Ausonius Epist XXV 34, Apollinaris Sidonius Ep III xiii 10, Carm V 288 & Carm XXIV 29. Professor R. J. Tarrant notes that in Bion fr. 12 (Gow) there is a pairing of Theseus/Pirithous and Orestes/Pylades similar to what we find in Ovid.