^{page 19 note 1} Students who use Dr Postgate's collation should be warned that in noting the agreement or disagreement of the other MSS with L he habitually employs the letter 0 to mean FDVN, though throughout the rest of the pamphlet it regularly means (A)FDV as opposed to N.

^{page 19 note 2} Dr Postgate (p. 39) bases or essays to base a conjecture on an unsupported reading of L at III iv 22 ‘me sat erit sacra plaudere posse uia’ NV, media DF, uoc. om. L, where he proposes ‘me sacra sat erit.’

^{page 19 note 3} v has three of them, but that does not count as confirmation.

^{page 19 note 4} There are also two places where Dr Postgate (pp. 39–41) builds conjectures, as he is quite entitled to do, on the joint testimony of F and L; but the conjectures themselves appear to have no advantage over the proposals of earlier critics. At III xii 43 ‘Sirenum sundo remige adisse lacus’ F has latus (s in ras.), L latreus, Dr Postgate accepts latus and alters Sircnum to Sicanium: if the vulgate needs changing, this change is no easier than Schrader's locos or than transposing lacus with the domos of the hexameter. At III xvii 17 ‘dum modo purpureo spument mihi dolia musto’ spument is an old correction, DVN have numen, F numine, L numerē, Dr Postgate offers cumulem which is not likely to oust the vulgate : st and sc are both confused with n, and the corruption of sp is only a trifle harder; then the archetype will have had nument, altered on the one hand to numen, on the other to numene whence numine and numerē.

^{page 20 note 1} To show my meaning I take parallels from N when at II xxxii 8 DV give rightly tibi me, N time, F timeo, we judge from comparing DV with F that N has here kept the corrupt reading of Φ while F has corrupted it further, just as we judged at III xix 6 that L had kept and F corrupted the fontes of Φ but when at III vi 41 FDV give quid mihi si and N quod nisi et we do not infer that N preserves the reading of Φ or 0, since FDV consent against and N had elsewhere to draw from.

^{page 21 note 1} Dr Postgate on the contrary says (p. 66) ‘we have seen that the concurrent testimony of L and N as to the reading of the family outweighs the dissent of F.’ I suppose the reference is to p. 33 where he quotes examples of agreement in spelling between N and L which he thinks must have been in Φ because ‘it is most improbable that correctors of this period would have troubled about trifles like equm, tinguere, pignera, murrea and so forth.’ The example of f and v would alone suffice to show that correctors write in the margin many readings which are in no sense corrections but merely variants, often insignificant, sometimes senseless, which have caught their eye in other MSS : then the next copyist stolidly incorporates them in his text. But it too often happens that scholars, instead of acquiring by observation a knowledge of what scribes were, prefer to frame from considerations of probability a notion of what scribes must have been.

^{page 21 note 2} Dr Postgate occupies two pages, 35–37, in demonstrating what he calls the honesty of L. He finds that L contains few readings which can be imputed to the conjecture of its scribe; he compares D, which Baehrens and I have praised for its honesty, and finds that D contains more readings of this sort; and he concludes ‘in honesty then it is clear that L is superior to D.’ Even if the term honesty is thus restricted, the amount of such conjecture in D is so small that this superiority of L's is evanescent; and an honesty which is compatible with such adulteration of the text as appears in L is not much to boast about.

^{page 22 note 1} Dr Postgate writes on p. 27 ‘Mr. Housman's advocacy of the value of F's isolated witness involves him in a curious inconsistency. He follows Baehrens in maintaining that A “is the most faithful representative of its family” Journal of Phil, xxii p. 99. It certainly then “happens curiously” that in the poems in which we have both A and F, A should give of itself but one true reading ‘solacia’ I v 27, which Mr Housman thinks after all may be an accident, and F three (or four), ib. p. 100 sq.' The only reason why I appear to Dr Postgate to be involved in a curious inconsistency is that he has forgotten the facts, which are these. A gives a far greater number of true readings than F, but wherever it gives a true reading that reading is also given either by F or by N or by both. F, which gives a far less number of true readings than A, gives two or three true readings which are given neither by A nor by N. I set three boys twelve sums : Tom does the first nine, Dick the first seven, and Harry the last eight; and I say Tom has done most, although every sum done by him has also been done by Dick or Harry or both, and although the three last sums have been done only by Harry; and I do not expect any one but Dr Postgate to tell mo that I am thus involved in a curious inconsistency.

^{page 23 note 1} Dr Postgate disputes the proposition that N has borrowed from Δ, but I shall come to that point presently.

^{page 23 note 2} I should add F, from II i 63 onwards, because it is there the only respectable representative of the family Φ; but with that exception I subscribe to Dr Postgate. I attach, little weight to F's unsupported readings from I i 1 to II i 63, or to the unsupported readings of D which I cite J.P. xxii pp. 101—3.

^{page 25 note 1} Baehrens, disliking N as Dr Postgate dislikes DV, promptly remarked ‘interpolate’: I in J.P. xxi p. 154 resisted him as I am here resisting Dr Postgate.

^{page 27 note 1} In the same paragraph I am told, with a compliment to soothe my vanity, that at II xxviii 9 I have not sufficiently regarded manuscript abbreviations, because, the true reading being peraeque, I quoted N's per aequae as an example of its superiority to the other MSS, of which D has (and V probably had) pareque and F paremque. ‘But’ says Dr Postgate ‘the original of the readings of all our MSS is neither “pereque” nor “pareque” but “peque” which L presents.” It pleases Dr Postgate to say so, but the statement has no other ground. That the readings of DVF are due to քeque, which their common parent 0 had misinterpreted as pareque, is probable; but 1 do not know what Dr Postgate means by saying that քeque was the original of N's per aequae : N, as he is aware, derives scores of readings from an older source than 0, and there is not a hint that this source had քeque rather than peraeque. The քeque of L no more tells for that opinion than the pere¸que of v for the contrary. But suppose I concede the point, what follows? that ‘N is here no better than D, V or F [Dr Postgate does not really mean “or F”]; for it has misdividcd the word, while they have wrongly expanded the abbreviation.’ Then N is better, because truth is more obscured by wrongly expanding the abbreviation than by misdividing the word. It will be observed that Dr Postgate's zeal for N has here succumbed to his tenderness for L.

On p. 23 an equally baseless charge of neglecting abbreviations is brought against Mr Leo and supported only by flat contradiction.