7 MSS: A (30r) Marcabruns, C (172r-v) Marcabru (C Reg. marc e bru), E (153) Mareabra, I(113v) Marcabrus, K (104v) Marcabrus, r(204v-205r) Marcabrun, rf (304v-305r) Marcabrus.

Analysis of the manuscripts

The xMSS divide between AIK, CE and T. AIK lack 42-47 and 55-60, and have common errors in 48 and 49. CE lack 19–21. AIK and CE also have minor divergences in 12, 18, 25 and 30. A few minor anomalous agreements across the groups (9 IC sabetz, AKE sabretz; 27 the presence of the article l in AET, absent in CIK) are probably due to convergence (see Roncaglia, ‘Ilgap’ p. 53), There is almost certainly an archetype error in 47, and possibly another in 49.

A contains an isolated error in 21, and isolated readings in 11, 19 and 28. I has an individual error in 20 and an individual reading in 21; IK share readings in 4, 11, 13, and 17, and IChas an isolated error in 27. Chas individual readings in 12, 39, 46, 47, 51, with isolated errors in 33 and 40–41; it may be suspected of cleaning up difficulties, for example in 12 to avoid repetition of sab(r)etz. E is individual in 25, 32, 46, 52, 54 (where it is unlikely to be correct against the joint testimony of all the other MSS), and contains individual errors in 4, 9, 40–41, 42 and 50.

T has too many errors and idiosyncrasies to be chosen as base, and in no case is it certain that Toffers the only correct reading against a common error in ACEIK; its few interesting variants, which we signal in the notes, are insufficient to warrant the space of a separate edition. For further discussion see Roncaglia, ‘II gap’, pp. 53-59.

Roncaglia rejects CE on the basis of his analysis of 11–12, 40–41, and 49. In the case of 11–12 he treats CET as a block (’Il gap’, p. 57), objecting to an awkward repetition of si in C8-9, the clash between 2 p. and 3 p. verbs in 11–12 of Ef and the similar clash which would have resulted between 8 and 9 in T if the verb had not actually been omitted in 9.