Chanson dc Roland. 121 



We must acknowledge, however, that we have no contirmator}- 

 evidence of other than technical nature, ^ as we had for the three 

 technical flaws cited on pp. 115-116. We must be content to reiter- 

 ate our feeling that the two strophes in question are very dissonant 

 with the greater part of O, and to point to the fact that the second, 

 at least, is quite consonant with the habits of mediaeval cop}'ists : 

 suppose a scribe desirous of currying favor with some descendant 

 of Gualtier, or a retoucher of logical bent, worried by the promin- 

 ence of Roland's liegeman at the end of the battle, and the pres- 

 ence of vv. 803-813 in O is explained. 



But beside these minor imperfections there is one of much greater 

 consequence : vv. 792-798. These lines are the climactic lines of 

 the chapter, for in them we learn that the Twelve Peers are to 

 remain in Roncesvalles. Yet only eight of them are named; and, 

 what is worse, two comparatively unimportant names are inter- 

 mingled. The climactic effect is missed. But it is evident, in m}' 

 opinion, that the original list of vv. 792-798 corresponded exactly 

 to that of vv. 2402-2409 ; for compare v. 794 with v. 2404, v. 795 

 with V. 2405, vv. 796-797 with vv. 2408-2409. The faulty list 

 is sural}' due to the carelessness of a copyist. ^ 



The strophes of the Oxford manuscript are as well composed as 

 the chapters. With very few exceptions, they possess unity, cohe- 

 rence, emphasis. With very few exceptions, they are perfect not 

 only as units, but also as component units. Illustration would be 

 futile ; let us discuss the exceptions alone. The first is vv. 1404- 

 1411. But this strophe is imputable to a copyist.^ The second is 

 vv. 1593-1627. That a copyist has meddled here has been the 

 opinion of almost all Roland editors.* The last is vv. 2083-2098. 

 Here our position is similar to that which we took above in regard 

 to vv. 803-813; we feel it is very improbable that a strophe with 

 three such miserable lines as 2090, 2096, 2098 is genuine ^ ; and 

 we note that vv. 2095-2098 are of the kind to which mediaeval 

 copyists were prone. 



1 I hesitate to accept the arguments of Donges, 1. c, p. 39, Note 65. 

 But the mere fact that lie suspects vv. 725— 73G, is perhaps not without 

 value as a confirmation of my view. 



- If I interpret correctly Miiller's remark on p. 75 of his edition (1878). 

 lie held tliis view. 



3 Cf. supra, pp. 115-116. 



* Cf. Mullers edition (1878), p. 160; and Stengel's edition, pp. 170-171. 



■■^ Cf. mfra, p. 124. 



