^{page 337 note 2} Bentley took 136: Haupt, 156: Vahlen, 158. These figures are from Kukula, De Codice Cruquii Vetustissimo p. 69. The same writer proves that the number of readings, which can with certainty be attributed to V, is far smaller than is usually supposed.

^{page 338 note 1} M is Mellicensis, at Melk near Vienna. It is collated throughout in the Epilegomena.

^{page 338 note 2} Keller here calls in β″, the supposed archetype of three MSS. of which β (tenth century) is the chief. This MS. is not elsewhere quoted. The analysis given above is only intended to show the main fluctuations of the MSS. The exact point at which a lacuna occurs, or at which a MS. changes its Class, is of noimmediate importance.

^{page 339 note 1} The table at the end of the Epilegomena is said to contain entries under 676 passages. Various statistics are compiled from it, but all of them leave a large balance of the 676 unaccounted for. We are told, among other things, that Class I agrees with Class II in 421 places, and with Class III in 121 places. Of these 542 places, it is right in 509 and wrong in 33. But we were told just before that in the whole of Horace Class I has 510 right readings, 100 wrong and 21 half-wrong. It would seem then that of the readings distinctive of the Class only one is right, but still this is the best class!

^{page 339 note 2} Keller would not admit that E is of Class III, but at least it is not Of Class I.

^{page 339 note 3} ν is wanting in the remaining examples.