^{page 12 note 1} But R and T, the descendants of A^A, seem to me to be more distant cousins than the descendants of B^A and C^A respectively.

^{page 12 note 2} Lindsay thinks that C^A was the ‘vulgate’ text, which T. G. Set himself to emend, the result being the B^A text (Ancient Editions of Martial, p. 55).

^{page 12 note 3} I have no figures, but if I had, I doubt if they would reveal anything of interest.

^{page 12 note 4} Viz. 1. 21. 7, 53. 4, 7, 55. 6, 61. 3, 66. n, 69. 1, 70. 5, 78. 7, 99. 12, 17, 109. 8, 116. 1; 2 Ep. 9, 6. 14, 14. 15,16. 4, 40. 5, 83. 3; 3. 2. 4, 6. 2, 58. 43, 77. 3; 4. 56. 5,63. 4, 66. 5,82.4; 5. 3. 6,15. 3,24. 8,45. 2, 75. 1, 78. 24; 6. 61. 10, 64. 32, 66. 6, 83. 5; 7. 7. 5, 14. 4, 27. 5, 87. 10,96. 4; 8. 3. 19, 50. 13, 59. 11,61. 4, 64. 6, 69. 1, 75. I, 3; 9. 12. 3, 30. 5, 45. 2, 51. 7, 57. 4, 58. 5; 10. 5. 4, 7. 9, 24. 11, 30. 1, 71. 4, 78. 8, 82. 7; 11. 1. 9, 3. 10, 15. 13, 37. 1, 3, 81. 4, 99. 2, 104. 21; 12. 32. 7, 16, 44. 1, 48. 7, 48. 15, 62. 15, 65. 8, 72. 3, 6, 77. 2, 84. 4; 13. 57. 2, 90. 1, 119. 1, 135. 1; 14. 24. 1.

^{page 13 note 1} Unless we are to take -que as = atque etiam, which is surely impossible in Martial.

^{page 13 note 2} The others are decus, nomen, spes, salus, tutela, levamen, pretium, cura, culpa, dolor, timor, tremor, lormentum.

^{page 13 note 3} The only plural abstracts thus used by M. are deliciance, amores, gaudia, lusus, opes and oscula.

^{page 14 note 1} Cf. 4. 14. 7.

^{page 14 note 2} Heraeus may of course nevertheless be right in saying that Quae est is itself a corruption.

^{page 14 note 3} e.g. at 9. 79. 5, 6.

^{page 14 note 4} It would perhaps ease dives to take sis as concessive, cf. 3. 5. 8, 5. 15. 6, but this is a much rarer construction in M. In that case the 2nd person would not be generic but refer to Milichus.

^{page 15 note 1} I discount Sp. 28. 10 as certainly corrupt.

^{page 15 note 2} And see Lindsay, , Ancient Editions of Martial, p. 60Google Scholar.

^{page 15 note 3} Cf. note 3 on p. 16.

^{page 16 note 1} For the abrupt haec sunt cf. 10. 104. 16; and for the si qua construction cf. 1. 34. 4.

^{page 16 note 2} Juv. 12. 24. ecce … audi is not quite parallel.

^{page 16 note 3} For stopgaps in our Mss. see 1. 103. 7, 114. 5; 4.89.6; 5. 4. 1; 6. 10.4; 8. 27. 2512. 17. 3; 14. 46. 2, and Housman's emendations of 2.36.3 and 8. 46 4.

^{page 16 note 4} Nor of ‘when’ (except possibly at 1. 53. 9).

^{page 17 note 1} Alternatively the homoeoteleuton 1 and 4 may have caused the trouble; i.e. the scribe went on after 4 with 5 and 6, then back to 2 and 3, and then his signal of restoration was misinterpreted, cf. B^A's omission of two lines at 7. 10. 12, 13; also 2. 7. 6–7; 1, 109. 15; 6. 42. 14; 8. 33. 12; 7. 92. 5; 11. 32. 1, 45. 3; 12 Ep. 13; 14. 25, 35, 98,162, and perhaps 13. 38, omitted by B^A owing to the like-ness of -antibus hedis to -ondibus horti, and 8. 27. 2, where R's Gaure is due to the likeness of (locup)leli and tibi.

^{page 17 note 2} As seems to have occurred at 12. 14. 3-4; see p. 22.

^{page 17 note 3} Tumores and furores were presumably punishments for lies, which he calls down upon himself if what he says is false. For the first cf. the Greek notion that blisters on the tongue were caused by lying.

^{page 18 note 1} For losses through homoeoteleuton see note 1 on p. 17.

^{page 18 note 2} I see that Friedrich, G. suggested taking Qui as relative to frater, as I do (Philol. 1909 p. 98)Google Scholar, but he then prients I. 3 without stops and without explanation. (No one seems to have noticed his repunctuation of 1. 7: units, cum sitis duo, C., sedetis? This, with Markland's sedebis, seems to me to be right.)

^{page 19 note 1} Cf. Munro's hypothesis of an omitted line at 2. 73, which he supplies most ingeniously. For other cases of homeoarchon cf. 1.53 (T's omission of two lines), 88. 4; 7. 31. 1059. 27. 3; 10. 76. 4; 11. 32. 1; 12. 55. 5, 10, 59. 8; 14. 10, 95 (omitted by B^A), and especially 7. 45. 11, which B^A has omitted from its likeness to the start of the previous line but one.

^{page 19 note 2} For Atlas = a dwarf the edd. compare Juv. 8. 32.

^{page 19 note 3} Housman (on Juv. 14 269) would read similem, but his only parallel is 12. 31. 6, where similes may well mean ‘like to one another’, not ‘like in colour to the dove-cot’.

^{page 20 note 1} si quid manias can mean nothing else, for, apart from its sense of ‘assign’, M. does not use the word in any other sense; cf. 104. 8.

^{page 22 note 1} The point of 11. 7–10 then would be that M. dares Publius to give up having lovely slaves and take ugly ones. But of course he will not.

^{page 22 note 2} Because Domitian is dead.

^{page 23 note 1} For omission or transposition through homoearchon, see note 1 on p. 19, and through homoeoteleuton note 1 on p. 17.

^{page 23 note 2} M. uses iste for ille and even hic, but not vice versa (haec aures at 7. 26. 4 is clearly corrupt).

^{page 24 note 1} But in any case there seems to have been a recognized body of title-headings by A.D. 400, if not earlier (see Lindsay, , loc. cit., pp. 53–4)Google Scholar .