1. Gnomon 54 (1926) 497–532Google Scholar = Beiträge II. 267–328Google Scholar.

2. ‘Bei einem Forscher, der Sprache und Stil so meisterhaft beherrscht und dem so viel Gescheites einfällt, muss es als ein Zeichen strenger Selbstzucht gelten, wenn er eigene Konjekturen nur ganz sparsam vorträgt’ (Fraenkel, 509 = 281). He put some twenty in the text, along with a few deletions, transpositions, and verses added exempli causa, naturally with no claim to reproduce the actual missing words.

3. Cf. Pasquali, , Storia della tradizione etc. (1934) 432Google Scholar Anm.: ‘Lucano ha bisogno urgente di un suo Knoche’ (quoted approvingly by Rutz, W., Lustrum 9 (1964) 245)Google Scholar. What did Knoche do for Juvenal's texO. See JRS 43 (1953) 223–4Google Scholar. Consider the guarded but penetrating remarks by L. Håkanson, whose recent untimely death has robbed us of one of the best critical minds in contemporary scholarship, on Gotoff, H.C.'s Transmission of the text of Lucan in the ninth century (1971) in PCPS 205 (1979) 26–29Google Scholar. Håkanson draws the conclusion that ‘it is questionable whether the constitution of Lucan's text will gain very much from Gotoff's work’; similarly C. O. Brink of Horace's uncollated codices; ‘while Textgeschichte is likely to profit from a much more extended study, the text of Horace is not likely to profit.’ And now Luck, who at one time took a sanguine view of the latent possibilities, opines in the preface to his edition of 1985 (p. 10): ‘die unerforschten Handschriften werden sicher unser Bild der Überlieferung bereichern, wenn auch für den Text selbst keine grossen Überraschungen zu erwarten sind.’ Then let them rest in peace. ‘The only object of seeking and collating manuscripts is to restore the author's text, to recover what he said’ (Housman, in the first paragraph of his preface).

4. For instance Lipsius' neglected palmary pullato in 2.480 oppositus quondam pollulo tiro Miloni. Lucan was hardly likely to abuse Milo, whose killing of Clodius was warmly approved by his hero Cato. On the other hand pullato gives the reader the pointer he needs; cf. 1.323 Pompeiana reum clauserunt signa Milonem.

5. A feature of Luck's edition.

6. I am indebted to him, and to a reader who remains anonymous, for some useful criticisms.

7. tangere = ‘smear on (ointment etc.)’ is unattested. I am not entirely sure what to make of OLD's entry tango 3b ‘to touch (a substance) so as to cause some of it to adhere’, but in the two passages cited from Scribonius Largus 10 and 71 (as in 20, 71 again, 130, 227, and 240) the meaning is certainly ‘touch’, not ‘apply.’ Also unconvincing is A. Ollfors' explanation of tecto as ‘deckend aufgestrichen’ (Textcritische und interpretatorische Beiträge zu Lukan (1967) 19–21Google Scholar). In Copa 36 anne coronato uis lapide ista tegi take ista as ista bona (‘those limbs’ Fairclough).

8. See PCPS 28 (1982) 92–93Google Scholar. R. Mayer (ibid. 29 (1983) 54), while agreeing that that aquae with amnis is otiose, thinks it justified as a reminiscence of Virg., Aen. 7 464fGoogle Scholar. aquai … amnis. Virgil refers to water boiling in a cauldron, and cauldrons, unlike terrestrial rivers, have been known to contain liquids other than water, aqua (from aquā) has as good support in Lucan's manuscripts as aquae, if not better.

9. I cannot accept a defence of Afer, also suggested by Mr Morgan, as ‘“an inhabitant of the district Africa,” which roughly corresponds to the modern Tunisia.’ Since Africa was the name of the Roman province, it was naturally sometimes used to mean that particular region, but I know of no passage where its inhabitants are distinguished as Afer or Afri from other African peoples. In 5.765, the only other occurrence in Lucan, vagus Afer is clearly generic. And it is a fair presumption that Sidonius found Arzux in Lucan along with the other names on his list (he will have taken Psyllus from Book 9).

10. Or ‘child,’ if nepotis (G and some editors; not recorded by Housman) be preferred.

11. Unless Galba of 8.469 is distinguished from Galba of 10.194; but he does not have to be.

12. I have to thank my colleague, Professor A. Henrichs, for some helpful discussion of this note.

13. ‘Pharsalus’ as the title of the battle, which Postgate in his edition of Book 8 (p. xcviii) called an error both gross and gratuitous, is still with us; and so despite Housman is ‘Pharsalia’ as the title of the poem.

14. Or for that matter ‘if so.’

15. So stated in my note.

16. Håkanson, 's article ‘Homoeoteleuton in Latin poetry’ (HSCP 86 (1982) 87–115)Google Scholar gives some interesting statistics, but more detailed classification is needed.