Astronomicon

contents of 539–560. But between 560 and 561 M has de circulis caelestibus : de coluris on the celestial circles: on the colures. The former title is right; it states the contents of 561–804. The latter, de coluris on the colures, is wrong; the right place for it would be before 603 sunt duo there are two, and it states the contents of 603–630. But the verses between 563 and 567 had been lost, and the transposition of leaves had carried away 568–611, so that 612–630, describing the colures, followed on 563 (with only the intervention of 567, which this corrector was removing), and the colures were therefore the first circles occurring after 561. The corrector accordingly wrote de coluris before that verse, beside the older title de circulis caelestibus. In β both were reproduced; but α took de coluris to be a correction of de circulis caelestibus, and so L gives de coluris only. (In G there is no title and no space for it; de coluris is added by a later hand in the margin.)

Between 529 and 566 L and M have the title de parallelis circulis on the parallel circles, and G has it in the margin, apparently from the first hand. It has no business to follow 529, but it is appropriate to 566–602, and was probably added by the corrector together with his own verse 566 when he inserted 567 at this place.

The transpositions in book I are thus seen to have been earlier than the archetype. Another, the great transposition of III 400–IV 9 and IV 10–313, is later, and found only in MSS of the family α. The verses IV 10–313 are 304 in number, that is $38 \times 8$; and III 400–IV 9, together with the 9 titles, occupying 10 lines, which existed in this portion of the archetype, are nearly the same, 302. It looks therefore as if a quaternio a gathering of four sheets or eight leavesMr Garrod p. xix saw that the transposition must have been later than the archetype, and found the number 38; but beside overlooking the titles he went astray in his arithmetic and operated with ternios. of 38 lines to the page had here been folded inside out, so that its last half came first, as I suggested in my note at III 399. This suspicion I can now confirm. In L, our best representative of α, the number of lines in the preceding part of the poem, I 1–III 399, verses and titles and spaces for titles together, is 2364; and its text, as was pointed out by Bechert de Man. em. rat. p. 10 n. 9, begins on the 2nd leaf of its 1st quaternio, the 1st leaf, now lost, having contained no verses. If α resembled it in these features, and had 38 lines to the page, the number of lines preceding III 400 was about 2440 ($2364 + 76$); and $608 \times 4$ (4 quaternios) is 2432. It becomes pretty clear that III 400–IV 313 were the contents of the 5th quaternio of α.

But Bechert p. 23 and Mr Thielscher Philol. LXVI pp. 131 sq. refer this transposition, though only one family is affected by it, to their imaginary archetype. For Bechert the interchanged