Page:The Library, volume 5, series 3.djvu/203

 MIRACLE CYCLES. 191 shows that each scribe generally pleased his own fancy in these matters, and consequently that to take them into account is only to obscure the evidence. I may also state at once that I regard it as proved that none of the cyclic manuscripts are compilations, and consequently that P is not an ancestor of any of the other texts. A few broad fa<5ts soon emerge. All the col- leftive manuscripts being dated, the direction of possible copying is known. Further, each contains at least one omission peculiar to itself, which proves that it cannot be the parent of any of the other texts, the possibility of insertion being nega- tived by the presence of the passage in P. Thus the general nature of the relation between the texts becomes apparent. To begin with, I will take a few of the more striking variants, and see to what detailed relation between the manuscripts they appear to point. The only omissions common to two texts are a number which occur both in W and K. These might point to omissions in their common original, which we will call F, but I am more inclined to ascribe them to mutilations in that manuscript. It is noticeable that in the neighbourhood of the common omissions there sometimes occur further omissions peculiar to K, a fa<5l most readily ex- plained by supposing a progressive deterioration of the original between 1592 and 1600. The evidence of transposition confirms the existence of F, but is of value chiefly in establishing a common ancestor for the cyclic manuscripts apart from P. There are, namely, four lines which appear in P