Instead of Roland (A 1793) it is Naymes who speaks first in E and F, 2570.
These few instances, the number of which might easily be increased, will certainly suffice to show the impossibility of regarding E as the original of A.
Only a short passage of the Didot MS. has been hitherto printed;[38] therefore the arguments drawn from a comparison of A with that printed passage cannot be considered as altogether irrefutable and final. But as the Didot MS. belongs to the same family of MSS. as E, we may at once presume, that as E cannot be taken for the original of A, the possibility of the Didot MS. being the source of A, is not very strong. Besides it may be stated, that no trace of the two additional lines (ll. 19 and 20[39]) which the Didot MS. inserts after l. 63 of a (or F) is found in A, although this version gives, in ll. 52 ss., a pretty close translation of the corresponding passage in F (ll. 50 et seq.). This may lead us to conclude that the Didot MS. was not the source of A.
Comparing now A with what is known of the Hanover MS. of Fierabras,[40] we find A resembling to H in the following names: Lucafer (only once Lukefer in A 2204), Maragounde (once Marigounde, A 1364), Maubyn A = Maupyn H.—A 1700 and 2831, which differ from F, equally agree with H. In the last case A agrees also with E (although differing from F). Now as we know that H together with D and E are derived from the same group z,[41] we may perhaps be justified in regarding a MS. of the latter group as the original of A. But a more detailed comparison of A with H being impossible at present, this argumentation wants confirmation.
The impossibility of regarding the Provençal version as the source [‹xxi›] of the Ashmolean Ferumbras, is proved by the fact that the long additional account, the ‘episode’ as Professor Grœber calls it,[42] is wanting in A. Another proof is given by A, ll. 5763 et seq., where A agrees with F, but widely differs from P.[43]
It seems superfluous to point out the inadmissibility of regarding the French prose version as the original of A, the first edition of the prose version being of a much later date than the Ashmole Ferumbras. But also that version from which the prose romance has been copied or compiled, cannot have been the original of A. For although the phrase of A, 3888—“A skuntede as a bore”—seems to contain some resemblance of expression with the reading of the prose Fierabras—“il commença à escumer come s’il fust ung senglier eschaufé,” which Caxton translates—“he began to scumme at the mouthe lyke a bore enchaffed”—the reading of A, ll. 1307 ss., which greatly varies from Caxton’s version (a translation of the French prose Fierabras), renders inadmissible the supposition that the original of the French prose version is the source of A.[44]
Having thus compared the Ashmolean Ferumbras, as far as can be done at present, with all existing versions of this romance, we arrive at the following conclusions.
The Ashmole Ferumbras is a pretty close translation of some French version, which we are at present unable to identify. Its original was neither of the same family (w) as the Fierabras, edited by MM. Krœber and Servois, nor yet of that of the Escorial version. Nevertheless, the original of Sir Ferumbras cannot have differed much from the common original, from which these two groups of MSS. are derived. To this original, called y by Grœber, the MS., from which A has been copied, appears to have been more closely related than to the Provençal version, from which it certainly is not derived. As the liberties which the author of Sir Ferumbras took in translating his original, consist only in very slight modifications, we may conclude [‹xxii›] from his closeness of translation in general, that in those passages of A which exhibit significant deviations from the known French versions, these variations are not due to the composer of the Ashmolean poem, but were already to be found in its original. Therefore the Ashmole Ferumbras may be considered as representing by itself the translation of an independent French MS., which perhaps belonged, or at least was nearly related, to the type y.
I now come to the consideration of the Sowdan of Babylone, which the simple analysis given by Ellis,[45] shows to be an essentially different work from the Ashmolean Ferumbras. Indeed, whilst the Syr Ferumbras represents only a portion (viz. the second part) of the original Fierabras [or Balan, as Gaston Paris has styled it],[46] the Sowdan approaches the original more nearly in that it contains the long ‘introductory account’.[47] For this first part of the Sowdan (as far as l. 970), although it cannot be considered as identical with the first portion of the old Balan romance, contains several facts, which, however abridged and modified, show a great resemblance with those which must have been the subject of the lost portion of the old original. Whereas the Ashmolean Ferumbras is, on the whole, a mere translation of a French original, the Sowdan must be looked upon as a free reproduction of the English redactor, who, though following his original as far as regards the course of events, modelled the matter given there according to his own genius, and thus came to compose an independent work of his own.
This point being fully treated in my Dissertation,[48] I need not again enter into discussion of it here. I only mention that the composer of the Sowdan has much shortened his original, omitting all episodes and secondary circumstances not necessarily connected with the principal action, so that this poem does not contain half the number of lines which his original had,[49] and that the proportion of the diffuse Ashmolean Ferumbras and the Sowdan is over five to one.[50] [‹xxiii›]