Page:The Library, volume 5, series 3.djvu/201

MIRACLE CYCLES. 189 has the banns, but omits seventy lines at the begin- ning, and adds at the end only the first six of Rogers' conclusion. Now it is clear that these banns, dating from the third quarter of the six- teenth century most likely, cannot have been in the archetype from which our cyclic manuscripts are derived. They might, so far as ascertainable dates are concerned, have occurred in the original of the group B D W K, but the sporadic manner in which they appear and the divergencies of the texts make it pretty certain that they were independently added in each case, and that they consequently throw no light upon the relations of the manu- scripts in which they occur. It is now time to consider the textual rela- tionship of our extant copies in greater detail. Deimling, who was the first to investigate the question, had no difficulty in showing on the one hand that H stood apart from the rest, and on the other that a peculiarly close bond united W and K. He was, of course, unable to place the missing D, but a few readings from that manuscript contri- buted by A. W. Pollard proved that it presented the peculiarities neither of H nor of W K. The obvious inference was that W and K have a common and exclusive ancestor, say F, that F and B (and possibly D) have an ancestor |3, and that only in the ancestor of j3 and H do we get back to the archetype of the extant manuscripts of the cycle. No scheme of the sort can, however, be regarded as satisfactory unless it will account not merely for a few selected readings or certain broad features of the texts, but for the whole body of minor variants