For _mollis_ used specifically of elegy (the _Aetia_ were in elegiac verse), see _EP_ III iv 85 and Prop I vii 19 (cited by André); for the word in an overtly Callimachean context, see Prop III i 19 '_mollia_, Pegasides, date uestro serta poetae'. _Tenere_ here has the sense 'keep to', as at _Met_ II 79 'ut ... uiam _teneas_' and Q Cic (?) _Pet_ 55 'perge _tenere_ istam uiam quam institisti [_Gruterus_: instituisti _codd_]'; Professor R. J. Tarrant rightly sees a suggestion of conscious artistic preference, and a faint allusion to the places where Augustan poets renounce the attractions of higher poetry. =33. TITYRON ANTIQVAS PASSERQVE REDIRET AD HERBAS= _B1C_. For the many variants and emendations proposed, see the apparatus. Housman has offered a defence of _B_ and _C_'s version of this line (937-39). He accepted Riese's printing of _Passer_ as a proper name ('M. Petronius Passer' is mentioned at Varro _RR_ III 2 2), and took the passage to mean 'He wrote bucolics, or, as Ovid puts it, he went back to Tityrus and the pastures of old': the construction is 'cum Passer rediret ad Tityron antiquasque herbas'. In writing the line, Ovid resorted to three devices, 'each of them legitimate, but not perhaps elsewhere assembled in a single verse'. The first is the delay of the preposition _ad_ after _Tityron_, which it governs; the second is the delay of _-que_, which properly belongs with _antiquas_; and the third is the placing of the verb between its two objects. For each of these devices Housman furnishes convincing parallels. Housman's argument is ingenious and informative, but I do not believe that he is right in defending the line: the accumulation of difficulties is suspicious, and the divergence of the manuscripts is greater here than at any other point in the book. Heinsius wrote of the line, 'haec nec Latina sunt, nec satis intelligo quid sibi uelint'. Like Heinsius, I believe the line to be deeply corrupted and, in the absence of further evidence, impossible to correct. =34. APTAQVE VENANTI GRATTIVS ARMA DARET.= Compare Grattius 23 'carmine et arma dabo et uenandi [_cod_: uenanti et _Vlitius_] persequar artis'. =34. GRATTIVS.= The manuscripts have GRATIVS (_CFLT_) or GRACIVS (_BMHI_); and _Gratius_ is what editors both of Ovid and Grattius printed until Buecheler pointed out (_RhM_ 35 [1880] 407) that _Grattius_ is the only form found in inscriptions, and is what is given in the oldest manuscript of Grattius, _Vindobonensis 277_ (saec viii/ix), which predates the manuscripts of _EP_ IV by at least four hundred years. =35. NAIADAS= _C. P. Jones_ NAIADAS A _HLI2_ NAYADES A _MT_ NAIDAS A _BCFI2_. Ovid elsewhere invariably uses the dative of agent with _amatus_ (_Am_ I v 12, II viii 12, III ix 55-56, _AA_ II 80, _Tr_ I vi 2, II 400, III i 42, IV x 40). As Professor Jones notes, following the interpolation of _a_, the shorter form _Naidas_ was introduced in _BCFI1_ to restore metre.