Chanson dc Roland. 125 



out of a total of 2927 ! Even though a more severe critic should 

 triple the list.i he would not be justified in contendincr that O lacks 

 consistent technical excellence of line. 



I hope that the foregoing exposition has rendered clear the nature 

 of the Oxford manuscript's technic. If so, I may consider m}- thesis 

 — that the original Chanson de Roland was a poem of marked and 

 consistent technical excellence — as proved.^ Let us turn to the dis- 

 cussion of several important Roland problems which may be satis- 

 factorily solved by means of it. 



* * * 



The tirst problem which confronts an editor of the Roland is of 

 course that of the interrelations of the manuscripts ; in other words, 

 the elaboration of a stemma. In the present writer's opinion, how- 

 ever, whoever attempts this for the Roland will find himself, sooner 

 or later, in an impasse, from which the usual methods of the study 

 of manuscripts will not help him to emerge. He will first discover 

 that no perfect stemma is here attainable. He will then narrow 

 the number of possible stemmata to two, the onl}^ two which have 

 resulted from a half-century and more of investigation on the part 

 of Roland scholars. But here he will find himself at a standstill, 

 for between the two there is no choice. They both show the same 

 number of flaws, and the exponent of either can explain the flaws 

 of his stemma by the theory that certain of the manuscripts are 

 ' Mischhandschriften,' or by supposing that some of the scribes who 

 wrote them knew the poem, at least portions of it, by heart.^ Un- 



^ It seems to me impossible that auy unjirejudiced critic should more 

 than triple the list. 



* Why should this thesis surprise us as much as it does ? In eleventh- 

 century France were generating the principles of Grothic architecture : is 

 it strange that a French writer of that time should possess a w^onderful 

 instinct for sti'ucture? 



' The complete bibliographj- of the manuscript problem is too extensive 

 to be given here : cf . Seelmann, Bibliographie dcs altfranzdsischen Rolands- 

 liedes (Heilbronn, 1888), pp. 73—77. Tollowing are the most prominent 

 exponents of one or the other stemma : (1) of the ' Oxford stemma ' (cf. 

 infra^ p. 126) : Ottmann, Die Stellung von V 4 in der Ubcrliefei-jmg dcs alt- 

 franzosischen Rolandsliedes, Heilbronn, 1879 ; Scholle, (a) Zeitschrift f. roma7t. 

 PhiloL. iv. 7—34 ; (b) ibid. 195—222 ; (c) Der Stajumbaum der altfranzosischen 

 -iind altiiordischen Uberliefern7igen des Rola?zdsliedes^ Berlin, 1889 ; Pakscher, 

 Ztir Krifik tind Geschichte des franzosischen Rolandsliedes, Berlin, 1885 ; Fass- 

 bender, Die franzosischen Rolandhandschriften in ihrem Verhdltnis zn einandcr 

 jind ziir Karlamagnussaga., Koln, 1887 ; and especially Miiller, (a) in the 



