In 1896 Rudolf Ehwald published his monograph _Kritische Beitrage zu Ovids Epistulae ex Ponto_. I am often indebted to Ehwald for references he has collected; my notes on i 15 _ad summam_ and xiii 48 _quos laus formandos est tibi magna datos_ could not have been written without the assistance of his monograph. This said, the fact remains that Ehwald's judgment and linguistic intuition were exceptionally poor. He had not relied on Korn's apparatus for his knowledge of _B_, but had collated it himself; and the intent of his monograph was to establish _B_'s authority as paramount. A typical example can be seen at ix 71. Here _FILT_ offer _cum ... uacabit_ and _MH_ have _ut ... uacabit_, while the reading of _B_ and _C_ is _quod uacabit_. In one of the examples Ehwald adduces, _Fast_ II 18, _uacat_ is found in only a few manuscripts, and it can easily be seen how it arose from _uacas_; all the other examples are instances of _quod superest_ or _quod reliquum est_. The cumulative effect of these examples is to demonstrate that _quod ... uacabit_ is not a possible reading. This insensitivity to the precise meaning of the passages he discusses is usual with Ehwald, and his book, although useful, is an extremely unsafe guide to the textual criticism of the poems. It has unfortunately exercised a decisive influence on all succeeding editions.

The first of these editions was Owen's 1915 Oxford Classical Text of the poems of exile. In the preface Owen acknowledges the influence of Ehwald: "adiumento primario erat R. Ehwaldi, doctrinae Ouidianae iudicis peritissimi, uere aureus libellus ... in quo excussis perpensisque codicibus poetaeque locutione ad perpendiculum exacta rectam Ponticarum edendarum normam uir doctus stabilire instituit' (viii). In most instances Owen follows Ehwald's recommendations, altering _in_ to _ab_ at i 9, _prospicerem_ to _aspicerem_ at ix 23, and at ix 44 abandoning Korn's _decretis_ for the manuscripts' _secretis_.

Owen's reliance on Ehwald was noticed by Housman (903-4) in his short and accurate review of Owen's edition: 'In the _ex Ponto_ Mr Owen had displayed less originality [than in his 1889 and 1894 editions of the _Tristia_] and consequently has less to repent of. Most of the changes in this edition are made in pursuance of orders issued by R. Ehwald in his _Kritische Beitrage_ of 1896; but let it be counted to Mr Owen for righteousness that at III.7.37 and IV.15.42 he has refused to execute the sanguinary mandates of his superior officer'.

As in Owen's earlier edition, the apparatus is so short as to be misleading. His choice of manuscripts is too small, and exaggerates the importance of _B_ and _C_; even of these two manuscripts his report is inadequate. At ix 73 he rightly prints _laxate_; the apparatus gives no indication that this is a conjecture, and that all manuscripts, including _B_ and _C_, read _iactate_, which he had printed in 1894. At xi 21, where _B_ gives _mihi_, indicated by Burman as the correct reading, Owen prints _tibi_ and does not mention the variant in the apparatus. The situation is naturally worse with readings of manuscripts other than _B_ and _C_, and with conjectures. In general, Owen's apparatus can be trusted neither as a report even of the principal readings of the few manuscripts he used, or as a register of critics'

views of the constitution of the text.

In the same year as Owen's second text there appeared at Budapest Geza Nemethy's commentary on the _Ex Ponto_, of which twenty-six pages are devoted to the fourth book. The notes are too sparse and elementary to form an adequate commentary, consisting largely of simple glosses. They are a useful supplement to a plain text of the poems, however, and Nemethy sometimes notices points missed by others: he correctly glosses _Augusti_ as "Tiberii imperatoris" at ix 70. The notes are based on Merkel's second edition; Nemethy lists in a preface his few departures from Merkel's text.

In 1922 Friedrich Levy published his first edition of the _Ex Ponto_ as part of a new Teubner edition of the works of Ovid. The apparatus was a reduced version of that prepared by Ehwald, 'Qui ut totus prelis subiceretur ... propter saeculi angustias fieri non potuit'. Levy's text is virtually identical to Owen's, but the apparatus is more complete.

It contains a full report of _B_ and _C_, and also of the thirteenth-century _Gothanus memb. II 121_. This last manuscript has the correct _cernet_ at ix 46, where most manuscripts read _credet_; but otherwise its readings are of very poor quality, consisting of simple misreadings (i 24 _magnificas_ for _munificas_, vii 30 _uento_ for _uenit_, viii 37 _habendus_ for _abunde_), simplified word order (vi 25 _tuas lacrimas pariter_ for _tuas pariter lacrimas_, xvi 39 _et iuuenes essent_ for _essent et iuuenes_), and intrusive glosses (viii 61 _captiuis_ for _superatis_, xvi 47 _me laedere_ for _proscindere_). The manuscript does not deserve the important place it has in the editions of Levy, Luck, and Andre[16]; Ehwald presumably included it in his apparatus because of its easy accessibility to him at Gotha, where he lived. No other manuscripts are regularly reported, so Levy's apparatus gives a false impression of the evidence for the text, although he often reports isolated readings from the manuscripts of Heinsius.

[Footnote 16: My knowledge of the manuscript is drawn from Andre's apparatus.]

Levy omitted conjectures 'quatenus falsae uel superuacuae uidebantur'; the result is that Korn's elegant _decretis_ does not appear even in the apparatus at ix 44, and the same fate befalls Scaliger's _coactus_ at xiii 27.

In 1924 the Loeb Classical Library published A. L. Wheeler's text and translation of the _Tristia_ and _Ex Ponto_. His text is based on Merkel's second edition, on Ehwald's _Beitrage_, and on Owen's Oxford Classical Text. In several places he rightly abandons _B_'s reading, printing _hanc_ for _ah_ at i 16 and _perstas_ for _praestas_ at x 83; at iv 45 he was clearly tempted to print Heinsius' _quamlibet_. His judgment is good, and if Ehwald and Owen had supplied him with more information on other manuscripts and on the Heinsius-Burman vulgate, his text might well have superseded all previous editions. His translation is accurate, and in corrupt passages indicates the awkwardness of the original; I have often quoted from it.

In 1938 there appeared the elaborate Paravia edition of F. W. Levy, who in the period following his earlier edition had altered his name to F.

W. Lenz. The text is virtually unchanged from his edition of 1922, but has a much larger apparatus, which includes a large number of conjectures omitted from the earlier edition; I am indebted to Lenz for many of the conjectures I report, particularly at xvi 33. The large size of the apparatus is, however, deceptive; most of the manuscripts he knew of only from the reports of Heinsius, Korn and Owen, and the reports are therefore incomplete: the only manuscripts reliably reported are _B_ and _C_. Since Lenz does not usually give the lemma for the variants reported, it is difficult to tell which manuscripts offer the reading in the text. Much space is wasted by reports of the readings of several heavily interpolated mediaeval florilegia; more is wasted by an undue attention to mediaeval spellings and attempts to reproduce abbreviations and to show the precise appearance of secondary corrections. These factors combine to render the apparatus virtually unreadable.

In 1963 Georg Luck published the Artemis edition of the _Tristia_ and _Ex Ponto_, with a German translation by Wilhelm Willige. Luck shows some independence from Lenz, at i 16 printing _hanc_ for _ah_, at iii 27 _sed et_ for _subito_, at viii 71 _mauis_ for _maius_, at viii 86 _distet_ for _distat_, at ix 73 _laxate_ for _iactate_, at xii 13 _producatur_ for _ut dicatur_, and at xiv 7 _muter_ for _mittar_, each time rightly. He suggests a new conjecture for the incurable xvi 33, and a new and possibly correct punctuation of xii 19. The apparatus is misleading, consisting of isolated readings from _B_ and _C_ and a small number of readings from other manuscripts. No indication is given that _hanc_ at i 16 or _pars_ at i 35 are found only in a few manuscripts, and not in _B_ or _C_. Luck criticizes modern editors for ignoring the discoveries of their predecessors, and rightly prints Heinsius'

_Gigantas_ (codd _-es_) at viii 59. However, he shows no direct knowledge of Heinsius' notes or of the Burman vulgate, making no mention of such readings as _Gete_ for _Getae_ at iii 52, _leuastis_ for _leuatis_ at vi 44, or _fouet_ for _mouet_ at xi 20. The oldest edition named in his apparatus is that of Riese.

In 1977 F. Della Corte published an Italian translation of the _Ex Ponto_ with an accompanying commentary, of which fifty-eight pages are devoted to the fourth book. Most of the commentary consists of extended paraphrase of the poems; I have found it of little assistance.

The most recent text of the _Ex Ponto_ is the 1977 Bude edition of Jacques Andre. His text is essentially that of Lenz, although at ix 23 he rightly prints _prospicerem_ instead of _B_'s _aspicerem_. There are a significant number of misprints in the text, apparatus, and notes, and other signs of carelessness as well.

Andre makes full reports of only four manuscripts in his apparatus, _B_, _C_, _T_, and _Gothanus membr. II 121_[17]. This is an inadequate sampling. _B_ and _C_ form a distinct group, and the Gotha manuscript is too corrupt to merit a central part in an apparatus. The result is that _T_ is the sole good representative of the vulgate class of manuscripts that is regularly cited.

[Footnote 17: He collated four other manuscripts, _M_, _Bernensis bibl.

munic. 478_, _Diuionensis bibl. munic. 497_, and _British Library Burney 220_, but gives their readings only occasionally.]

For knowledge of many of his secondary manuscripts, Andre seems to have depended on the edition of Lenz. Since much of Lenz's information was drawn from Heinsius and other earlier editors, this means that Andre is often giving unverified information from collations made more than three centuries previously. He did not realize that the Antwerp manuscript he collated (our _M_) was Heinsius' _codex Moreti_, whose readings Lenz sometimes reports; the result is that he reports the same manuscript twice, under the sigla _M_ and _N_.

At ix 127 he cites the sixth-century Wolfenbuttel fragment in support of the unassimilated spelling _adscite_ (the assimilated form _ascite_ is supported by the inscriptions and by the ancient manuscripts of Virgil).

In fact, the word is not found in the fragment, which preserves only the first three letters of the line.

Finally, Andre shows insufficient knowledge of the Heinsius-Burman vulgate; this is evident not only from the text but from the introduction, where he prefaces his list of principal editions by saying 'Nous ne mentionnerons que les editions fondees sur des principes scientifiques, dont la premiere est celle de R. Merkel, Berlin, 1854'

(the edition was published at Leipzig in 1853).

In spite of what I have said against it, Andre's edition has considerable merit. His apparatus is the first to supply a lemma for each variant reading reported, and is clear and easy to read. His selection of manuscripts is inadequate, but at least he makes a full report of the four manuscripts he uses. The apparatus is in every way a great improvement on that of Lenz. At the same time, he provides a clear prose translation, an informative introduction, ample footnotes, and thirteen pages of "notes complementaires". His notes sometimes come close to forming a true commentary, and I often quote from them.

In preparing this edition of the fourth book of the _Ex Ponto_, I have carefully read all the editions discussed above, and have attempted to include a comprehensive list of conjectures in the apparatus. I have read Burman's variorum edition with particular attention, and have often restored readings favoured by Heinsius to the text. A complete examination of the manuscripts must await a full edition of all four books of the _Ex Ponto_; but on the basis of published editions I have selected the nine manuscripts that appeared most likely to assist in establishing the text, and have included full reports of their readings in the critical apparatus. I believe that even this preliminary apparatus gives a clearer picture of the evidence for the text of _Ex Ponto_ IV than any previous edition.

P. OVIDI NASONIS

EPISTVLARM EX PONTO LIBER QVARTVS

CONSPECTVS SIGLORVM

_G_

Herzog August Bibliothek Wolfenbuttel: Cod. Guelf. 13.11 Aug. 4 (fragmentum Guelferbytanum) saec v/vi

continet ix 101-8 et 127-33, xii 15-19 et 41-44. uersus saepe non integri.

_B_

Monacensis lat. 384 saec xii

_C_

Monacensis lat. 19476 saec xii

_M_

Antuerpiensis Musei Plantiniani Denuce 68 saec xii/xiii codex Moreti Heinsianus

_F_

Francofortanus Barth 110 saec xiii

_H_

Holkhamicus 322, nunc British Library add. 49368 saec xiii

_I_

Laurentianus 36 32 saec xiii primus Mediceus Heinsii

_L_