^{page 67 note 1} Before ed. Mg, 1518, printed text omits 33 and stops at 40. 37. 3.

^{page 67 note 2} For Aganensis see below, p. 9.

^{page 67 note 3} Mg began 33. 17. 6 ‘eis partibus.’

^{page 67 note 4} Kreyssig, 1839 ed. of Book 33 (collated B for rest of decade).

^{page 67 note 5} The fragments consist of the following portions: 33. 34. 9—33. 35. 5, 33. 36. 5—33. 37. 6; 34. 29. 11—34. 29. 14, 34. 31. 19—34. 32. 2; 35. 5. 10—35. 6. 1, 35. 8. 4—35. 8. 9; 39. 36. 4—39. 36. 16, 39. 37. 1—39. 37. 16.

^{page 68 note 1} As regards Franciscus Modius (the third inspector of Mg), Drakenborch(Vol. VII., p. xlv) has shown the worthlessness of his work.

^{page 69 note 1} E.g., in 37. 2. 12 the text (before investigation of Mg) gave ‘binae eo anno decumae frumenti imperatae’; Carbachius gives us ‘binae eaeque proximae decumae imperatae.’ Surely Carbachius found ‘eo anno’ in Mg and also ‘frumenti’? He was wishing to comment on the fact that ‘eaeque proximae’ occurred after ‘binae,’ and he completed the sense of his quotation by writing ‘decumae imperatae.’ If this theory is not accepted, then we must suppose either with Drak. that ‘aeque proximae’ as a gloss (made by an arithmetical reader) replaced ‘eo anno’ in Mg, or with Madvig that B +dett. had 1‘eo’ ‘for’ aeque ac proximo, ‘while Mg omitted ‘anno.’ This would imply one omission in Mg and one in B + dett., which seems unlikely. But if we admit that Carbachius gave an incomplete quotation, Madvig's emendation is comparatively easy:

(Madv.) BINEAEQO'AC PROXIMO ANNO was corrupted into (say)

BINEAEQUEPROXIMOANNO, which became(say)

BINEAEQUEPROXIMEOANNO, which became(say)

BINEEAEQUEPROXIMBEOANNO, which, I think, Carbachius thought Mg read; cf. 35. 14. 7.

^{page 69 note 2} Since writing this, I have had access to Aduchs ‘de Gel. codice Spirensi.’ I find that the same point is made here and the relation between ‘S’ and the ‘dett’ demonstrated.

^{page 70 note 1} E.g., cf. the large number of readings in which Gelenius' statement of reading is sufficiently approximate to that of Carbachius to suggest that Gelenius was quoting from Mg rather than from any other MS. Where they are conspicuously different, then the reading of Gel. is not far removed from B—e.g. 37. 23. 10 (cf. elenchus). The hypothesis is thus confirmed.

^{page 70 note 2} Cf. Drak., p. 327, Vol. VII.

^{page 71 note 1} See below.

^{page 71 note 2} Henceforth ‘Mg’ will mean ‘Mg as quoted by Carbachius’ unless otherwise stated.

^{page 71 note 3} Cf. in elenchus: 34. 4. 3, 34. 16. 1, 34. 34. 2, 35. 2. 7, 35. 11. 11, 35. 20. 4, 36. 10. 1, 37. 16. 9, 37. 30. 7, 37. 41. 10, 38. 21. 6, 38. 25. 1, etc.

^{page 72 note 1} Cf. 34. 8. 2, 34. 31. 18, 34. 44. 3, 35. 8. 7, 35. 20. 3, 36. 7. 4, 36. 23. 10, 37. 18. 4, 37. 20. 4, 38. 23. 6, 38. 38. 8, etc.

^{page 72 note 2} So, too, on Drak.'s authority, four of his MSS.

^{page 72 note 3} It may be worth noticing that the errors inthese instances are of the type that occurred at a period when corruption according to sense was common.

^{page 72 note 4} For phrase cf., with Drakenborch, 32. 33 ‘in antiqui formulam iuris restituere.’

^{page 73 note 1} Zingerle cites B as reading ‘et,’ Kreyssig does not mention it.

^{page 73 note 2} As for the reading of Carbachius from Mg, it appears likely that miles had been written as a gloss in Mg or a predecessor to account for the singular ‘Romanus’; ‘praesidio’ may have crept into the text of Mg, when (‘miles’) ‘misesset’ had been introduced and a predicative dative seemed to be expected, or Carbachius himself may have misread what he saw.

^{page 73 note 3} Cf. 34. 2. 11, 34. 7. 2, 34, 22. 1, 35. 6. 6, 35. 26. 9, 35. 27. 2, 36. 14. 10, 36. 17. 5, 36. 19, 2, 37. 1. 6, 37. 11. 6, 37. 48. 4, 37. 48., 38. 45. 8, etc.

^{page 74 note 1} Cf. 34. 2. 2, 34. 13. 4, 34. 28. 8, 34. 41. 4, 36. 16. 1, 36. 33. 5, 37. 17. 10, 37. 21. 4, 38. 8. 10, etc.

^{page 74 note 2} Cf., however, 35.10.4, where Gelenius (‘antiquior lectio’) reads ‘petebant,’ B omits verb, and ‘tenebanl’ occurred in some of early printed texts, while others (with all Roman dett.) read the same as Gelenius, and four of Drakenborch's codices read ‘pendebant.’

^{page 74 note 3} E.g. 34. 1. 6, 34. 35. 4, 37. 20. 4.

^{page 74 note 4} It may be worth noting that in another passage where B leaves a blank space the syllable ‘con’ is in question. In the speech of Cato about repeal of Oppian Law B leaves a gap after ‘feminas’ (they were too much for him). All the examined ‘dett.’ give ‘concitandas,’ 2. 7; cf. also 37. 54. 28.

^{page 74 note 5} Cf. 34. 4. 18, 34. 19. 11, 37. 24. 3, 38. 2. 10.

^{page 74 note 6} Cf. 34. 22. 2, 34. 27. 10, 34. 58. 5, 35. 7. 2, 35. 39. 5, 35. 44. 1 (‘aliis’), 36. 1. 1, 37. 10. 5, 37. 13. 1, etc. The importance of this category for testing the soundness of Carbachius' and Gelenius' readings has already been suggested.

^{page 75 note 1} In the margin of the Agenensis is written ‘deest numerus’ by a hand that may be that of Valla.

^{page 75 note 2} E.g. 34. 46, 4, ‘locis apertis’ B dett.: ‘locis idoneis’ Mg (where one of Drak.'s MSS. gives ‘aptis’); and 35. 47. 2 (‘in naue’ B: ‘inane’ dett.: ‘uanum’ Mg) and 37. 32. 13 (‘potentiora’ B dett.: ‘opulentiora’ Mg.

^{page 75 note 3} E.g. 34. 44. 8, ‘consciorum’ majority of dett. ‘sociorum’ Mg (probably correction of barbarous word ‘consociorum’) and one late MS.: ‘consociorum’ B; and 35. 14. 7 (where Carbachius probably saw ‘ultimas’ in Mg, but did not quote it); cf. also 37. 7. 15, 37. 54. 28 (vid. sup.).

^{page 75 note 4} British Museum, Harl. 2493, described by ProfessorWalters, C. F. in Class. Rev. XVIII., p. 392Google Scholar.

^{page 75 note 5} An attempt to connect this ending with that of Gelenius' ‘S’ proved fruitless.

^{page 75 note 6} E.g. 36. 4. 2, AB ‘etiam’: X ‘et’; 36.3.7, ‘nuntiaret’ B: ‘nuntiare’ X: ‘nuntiare. et’ A; 36.43. 1 (cf. above); 37. 13. i, ‘praeteruehemtls’ AB; ‘praeteruehentes’ X; 38. 23. 6, ‘late romanis’ B: ‘later ols’ A: ‘late inter omnis’ X.