Chiuisou dc Roland. 135 



not fulfilled ; in the other redactions it is ; therefore the arrangement 

 of the latter is to be adopted. But is it necessary to interpret the 

 conversation of the Counts so literall}' ? It may be that their re- 

 marks were of general, not specific, import — " Turpin, a man of 

 God, fiohts well ; we, men of battle, should do likewise." The pos- 

 sibility of this interpretation prevents the contention that Miiller's 

 reason is imperative. But, if need be, our thesis ma}' again be thrown 

 into the balance — is not the clearness of O's chapter, so consonant 

 with the technical excellence of x', a strong argument against alter- 

 ing o?i 



So much for Miiller's infidelity to his formula. But his edition is 

 unsatisfactory for another reason — because he reconstructed, not x, 

 but X' ; which was, to use his own expression, " eine schon getriibte 

 Quelle." He, in common with most Roland scholars, did not con- 

 .sider the reconstruction of the Original poem to be feasible, for the 

 reason that the nature of that original was a matter of doubt. 

 Almost all students of the question have thought that between x 

 and x' intervened a long process of accretion. To quote Professor 

 Weeks : " To my mind, the process of development was so gradual 

 that, at no stage of the operation could one say : ' Here begins the 

 Oxford version.' " ^ No editor who holds this opinion would attempt 

 to reconstruct x. But, if my thesis be conceded, it must also be 

 conceded that between x and x', and between x' and O, there was 

 very little accretion, or indeed change of an}' kind^; that, moreover, 

 it is feasible to reconstruct x, merely by excluding from O what- 

 ever is due to its scribe, or to the scribe of x'. Here, again, the 

 thesis will be of use : never as sole reason for exclusion — for, as I 

 have already said, our Old French Homer may well have nodded 

 now and then — but often as a confirmatory reason, vv. 1406-1411, 

 for example, are attribvitable to some copyist on linguistic grounds 

 — technical considerations strengthen the case*; vv. 520-536 seem 



* Scliolle, Dcr Staiiiiubaiiiii . . . drs Rolaiidslicdes, pp. 22—23, argues for 

 the retention of O's order in this chapter. 



* Modem Language A'otes^ xxii, 191. 



* My thesis proves nothing in regard to the so7irccs from which the 

 poet drew his material. He may have compiled folk-songs ; he maj^ have 

 remade some story, as Tennj^'son remade the tale of the Red Book of 

 Hergest into his Geraint and Enid -^i he may have collaborated, as Bedier 

 would have it, with the monks of some abbey ; he may have spun lus . 

 own web. Such questions, to my mind, have nothing to do with the 

 reconsti'uction of the original text of the Chanson dc Roland. 



* Cf. snpiri, p. IIG, Note 1. 



