(a) and (b) mixed: melody-e, F. sb., dy-e, infin. mood, 675; espy-e, gerund, curtesy-e, F. sb., 795; hy-e, dat. adj., maistry-e, 841; dy-e, gerund, flatery-e, F. sb., 1063; curtesy-e, F. sb., hy-e, dat. case, pl. adj., 1251; dy-e, infin. mood, remedy-e, F. sb., 1479. Seventeen examples. (In all, fifty-five examples.)
Thus, in more than fifty cases, the Chaucerian habit is maintained, and there is no instance to the contrary. Even the least trained reader may now fairly begin to believe that there is some value in this proposed test, and may see one reason for supposing that fragment A may be genuine.
[§ 16]. A still closer examination of other rimes tends to confirm this. There are no Northern forms (as in B), no merely assonant rimes (as in B), nor any false or bad or un-Chaucerian rimes (as in both B and C), except such as can be accounted for. The last remark refers to the fact that the scribe or the printer of Thynne's edition frequently misspells words so as to obscure the rime, whereas they rime perfectly when properly spelt; a fact which tells remarkably in favour of the possible genuineness of the fragment. Thus, at l. 29, Thynne prints befal, and at l. 30, al. Both forms are wrong; read befalle, alle. Here Thynne has, however, preserved the rime by making a double mistake; as in several other places. A more important instance is at l. 249, where the Glasgow MS. has farede, herede, a bad rime; but Thynne correctly has ferde, herde, as in Chaucer, Cant. Ta. A 1371. So again, at ll. 499, 673, where the Glasgow MS. is right (except in putting herd for herde in l. 673).
At l. 505, there is a false rime; but it is clearly due to a misreading, as explained in the notes. A similar difficulty, at l. 1341, is explicable in the same way.
[§ 17]. So far, there is no reason why fragment A may not be Chaucer's; and the more closely we examine it, the more probable does this supposition become. Dr. Kaluza has noticed, for instance, that the style of translation in fragment A is distinctly better, clearer, and more accurate than in fragment B. I find also another significant fact, viz. that in my essay written to shew that 'the translation' is not Chaucer's (written at a time when I unfortunately
regarded the whole translation as being the work of one writer, a position which is no longer tenable), nearly all my arguments were drawn from certain peculiarities contained in fragments B and C, especially the former. I have therefore nothing, of any consequence, to retract; nor do I even now find that I made any serious mistake.
[§ 18]. The third result may, accordingly, be arrived at thus. Seeing that Chaucer really translated the 'Roman de la Rose,' and that three fragments of English translations have come down to us, of which two cannot be his, whilst the third may be, we may provisionally accept fragment A as genuine; and we find that, the more closely we examine it, the more probable does its genuineness become.
[§ 19]. Summary.—Having now discussed the three fragments A, B, C, successively and separately (though in a different order), we may conveniently sum up the three results as follows.
1. Fragment A appears to be a real portion of Chaucer's own translation. Its occurrence, at the beginning, is, after all, just what we should expect. The scribe or editor would naturally follow it as far as it was extant; and when it failed, would as naturally piece it out with any other translation or translations to which he could gain access. This fragment ceases suddenly, at the end of l. 1705, in the middle of an incomplete sentence. The junction with the succeeding portion is clumsily managed, for it falsely assumes that the previous sentence is complete, and leads off with a false rime.
2. Fragment B is obviously from some other source, and is at once dissociated from both the other fragments by the facts (a) that it was originally written in a Northumbrian dialect, though this is somewhat concealed by the manipulation of the spelling by a later scribe; (b) that it was written in a more diffuse style, the matter being expanded to the extent, on an average, of nearly twelve lines to ten; (c) that many licences appear in the rimes, which sometimes degenerate into mere assonances; and (d) that it is less exact and less correct in its method of rendering the original.