^{page 32 note 1} The two poems we call Dirae and Lydia are transmitted in our manuscripts as one poem, under the tide Dirae. They were rightly separated by F. Jacobs in 1792. It is, of course, with the former of them that I am concerned.

^{page 32 note 2} Fraenkel jusdy prefers a parenthesis in 61, dicat in 75, and raptorum in 82. But in 8, where he adopts the fifteenth-century conjecture dicat for the paradosis dicam, removing punctuation at the end of 7,1 follow him hesitantly. Mr Kenney retains dicam and punctuates thus: multa prius fient quam non mea libera auena: / montibus et siluis dicam tua facta, Lycurge. This is tolerable, for libera can convey the apposite sense ‘outspoken’. But undeniably the slight change to dicat effects much greater smoothness. How important this is in a poet not generally characterised by smoothness is hard to assess. Fraenkel further argues that ‘dicam would not be in keeping with the rest of the poem, where in corresponding expressions the plural is always used’, namely at 1, 2, 14, 26, 63, and 97. This argument may have some force, though there is no exacdy corresponding expression in the poem, and we find singulars at 66 per dam, 86 uisam, and 102 amabo.

^{page 33 note 1} At least this is one way of interpreting the line. The meaning of pius/impius is hard to grasp, and it may well be misleading to make impia here = ‘wicked’.

^{page 34 note 1} dulcius in 71 might also be included here, if the text is sound. I discuss the line later in this paper.

^{page 34 note 2} Anyone who doubts that this line is sentimental might, as Professor Jocelyn suggests to me, ask himself what real need these goats had for pity (miserae).

^{page 34 note 3} In 29 Rothstein kept the paradosis multa, compared Dirae 7, and punctuated thus: multa prius: uasto…ponto, etc. He may be right, though I agree with Shackleton Bailey that uasto…ponto would be ambiguous. Incidentally, as far as similarity of tone is concerned, it is immaterial whether Propertius is imitating the Dirae or vice versa.

^{page 34 note 4} And neither acceptable sense nor Latinity can be restored by minimal changes, of the kind favoured by some editors, such as that of dicens to dices (though this in itself is likely enough), combined with punctuational chicanery to produce things like non iterum dices ‘crebro “tua Lydia” dixu No sane persons write thus (not even third-rate poets), only certain editors of classical texts.

^{page 35 note 1} I have not overlooked the possibility that tua might refer to Iuppiter in 35–6, but I reject it, because, as I have said, our attention in 26 ff. is directed to the wood itself, and therefore to refer tua to Jupiter, who comes in importantly indeed but incidentally at 35–6, would create ambiguity and confusion.

^{page 35 note 2} One might further argue that the use of the same verb in dicens and dixti is, at the least, un-pleasing. But this argument would be invalid. The text may be corrected in such a way as to give the repetition both function and point, perhaps on the lines suggested by M. Schmidt's bold, but intelligent, quae for tua in 41.

^{page 36 note 1} This may well get some support from Lydia 4–7 uos nunc illa uidet, uobis mea Lydia lidit, / uos nunc alloquitur, uos nunc arridet ocellis, / et mea submissa meditatur carmina uoce, / cantat et interea, mihi quae cantabat in aurem.

^{page 36 note 2} The fifteenth-century conjecture nec adire for audire gives a sense, though a weak one. Haupt's haud ire seems to me questionable Latin, and its sense, if it has any, is similarly weak. Dr Diggle suggests laudare, comparing Virg. Georg. 2. 412 laudato ingentia rura. This gives a more pointed sense, and accords quite well with the laudatory tone of the following lines. But I do not think that the context indicates clearly that laudare is the idea required.

^{page 37 note 1} Note, for example, 7 auena, 15 auenas (different sense), and 19 auenis, or 13 ipsae non siluae frondes and 18 desini et siluis frondes, or impia at 31 and 45 as well as in the repeated formula of 3 and 62, or 81 crimine and 82 crimina. Of course, in a poem which makes such use of refrain one must be careful to distinguish between intentional and unconscious repetition. But there seems to be quite a lot of the latter.

^{page 38 note 1} Still, as Mr G. V. Wilkins reminds me, the nymphs who personify springs are not always castae. Where Diana was safe, Hylas was in peril.

^{page 38 note 2} Even if the phrase were a stopgap, that alone would not make it evidence of spuriousness.

^{page 38 note 3} The Pannonian veterans of A.D. 14, though their experience was rather different, might have appreciated this point. After twenty and more years of service they wanted agri. They got per nomen agrorum uligines paludum uel inculta montium (Tac. Ann. I. 17. 3).

^{page 38 note 4} Professor Jocelyn suspects the writer had in mind an etymology of ager whereby the word was supposed to convey the notion of fruitfulness. If this is so, the point here would be that, though in the word agelli itself there is promise of good fruit, it is only in the word, since because of the dirae the agelli will be infelices.

^{page 39 note 1} One may compare the vastly more elaborate treatment of a similar theme in Ovid, Trist. I. 3.

^{page 39 note 2} And it may have been written by the same poet in a different mood. I do not think it was, because of the extent of the stylistic difference between the two poems, but, in spite of that, the possibility of common authorship cannot as yet be excluded absolutely.

^{page 40 note 1} Of the conjectures Scaliger's tu nemus remains the most attractive.

^{page 41 note 1} The nearest he gets to imputing human feelings to nature is at 64–5 Battare, fluminibus tu nostros trade dolores: / nam nbi sunt fontes, tibi semper flumina arnica. The notion of woods etc. echoing the human voice (30 and perhaps 41) is hardly comparable, at least in the limited way this poet uses it.

^{page 41 note 2} tonderis (Putsch, Naeke) is tempting, but involves grave difficulties, namely (i) that, if it is present, from tondeo, it clashes unpleasantly with the futures which immediately follow, (ii) that, if it is future, it must come from tondo, a collateral very tenuously, indeed dubiously, attested in good Latin, and (iii) that, since the context suggests tonderis is future, normal word-formation that it is present, we are left uncertain how to take the word, but competent Latin poets studiously avoid placing their readers in such dilemmas.

^{page 41 note 3} The parenthesis in line 30, though a litde clumsy, is much less difficult, since its sense fits apdy enough with what precedes and what follows.

^{page 42 note 1} This may indeed be a manner of speaking borrowed from magic formulae.

^{page 42 note 2} The change is extremely slight, merely merito for merita. And for the writer to say that all the things he has cursed belong now to Dis is, in a poet, a likely sort of variant for such commonplace magical ideas as consecration dis manibus or dis inferis. It is because of the sense of merito, not that of omnia Ditis, that I regard Ribbeck's correction as uncertain. I find this bald and direct statement that the farm merits destruction is very incongruous in a poet who elsewhere handles this aspect of his theme rather obliquely (e.g. at 45–6 and 69–70). Of course one can see how he may have arrived at this thought: the farm is now occupied by an adueña, not by me, he is impius, and therefore the farm, now his, deserves to be destroyed, as he does. But this is to read a lot into a single word, merito. Still, for all this, the conjecture is very plausible.

^{page 42 note 3} If the same idea had been used earlier, as a refrain or otherwise, part of the case for removing 66 would vanish. As it is, the line's uniqueness heightens its incongruity.

^{page 42 note 4} This is what Ribbeck proposed, or rather one should say that he placed 66 before 82, since he also transposed 79 to stand in place of 66 and 80–1 to stand after 71. 1 think I have improved on his proposal but, if there is any credit in the transposition of 66, it is his, not mine.

^{page 42 note 5} I can see no cogent mechanical explanation of the omission, though the last two letters of the last two words of 83 and 66 are identical, namely -ia and -is.

^{page 43 note 1} One can scarcely hope for a consensus over the layout of this complex poem. A brief comparison of Mr Kenney's text with Fraenkel's and my own will indicate the main problems, and the different treatment of which they admit.

^{page 43 note 2} I particularly commend my text at the beginning of this paper to those who concern themselves with patterns and symmetries. They will find much in it to delight them.