Chanson de Roland, 127 



lengths of the Hnes x— x', x'— O, etc. Notice, then, that the stem- 

 mata show two radical differences : (1) in the one, O constitutes a 

 family in itself — in the other, O and V4 constitute a family ; (2) in 

 the one, O, of all the extant redactions, is by far the closest to 

 X' — in the other all the extant redactions are about equidistant 

 from x'. Because of these two differences I have called the one 

 the ' Oxford stemma,' the other the ' redactions stemma.' 



Now, if the original Chanson de Roland— iha.t is, x — was a poem 

 of marked and consistent technical excellence, a stemma which 

 results in the reconstruction of an x of poor technic is incorrect. 

 Let us then cease testing the two exasperating stemmata themselves, 

 and turn to the examination of their results ; that is, of reconstructions 

 based upon them. No one has ever attempted the reconstruction 

 of X— for the reason stated infra, p. 135 — but several scholars have 

 essayed that of x'. Their texts will serve our purpose as well as 

 wovild texts of X, inasmuch as x' — for reasons similar to those urged 

 in the case of x — was a poem of marked and almost consistent 

 technical excellence.^ I propose, then, to examine two reconstructions 

 of x', one of them based on the Oxford stemma, the other on the 

 redactions stemma : the second edition (1878) of Theodor Miiller, 

 and the edition of Edmund Stengel.^ I believe that I can show 

 that Miiller's text possesses marked and almost consistent technical 

 merit, that Stengel's possesses many technical imperfections — in 

 which event I shall have proved the correctness of the Oxford 

 stemma.3 



' Cf. siij>ra, p. 112. 



- Needless to say that these two are the most prominent of the manj?- 

 editions of the poem. 



* As I have already said {supra^ p. 112, Note 2), it is barely possible that 

 the technical excellence of O was due to some lost intermediary between 

 x' and it. In that case my arguments against' Stengel's stemma are 

 invalidated. But if Stengel's stemma — that is, the redactions stemma — is 

 correct, we have the unparalleled phenomenon of x', a poem of many 

 technical imperfections, having been reduced, merely by skiUful cutting, 

 to x'", a poem of marked and consistent technical excellence. Although 

 it is ti-ue that some— not many— mediaeval poems were improved by 

 retouchers, the phenomenon of a poem having been considerably short- 

 ened is, I think, unparalleled. Furthermore, Stengel's stemma, if correct, 

 involves several very improbable conclusions : for example, we must con- 

 clude that the scribe of O had the exceedingly good taste to leave intact 

 the greater part of x'", and, at the same time, the exceedingly bad taste 

 to reintroduce some of the worst technical imperfections of x', such as 



