Page:The Song of Roland.djvu/24

 lastly, there is the pre-eminent characteristic of the lines of this measure, each of which is strikingly “single-moulded,” as the word has been used of English—that is to say, held up at the end, and constructed all through so as to run to that end and stop. This arrangement is neither “blank”—that is to say disregardful of, and in fact shunning, any agreement of vowel sound at the end; nor rhymed—that is to say, constructed with couplet or some other arrangement so as to effect consonance of sound ending; nor stanzaed—that is to say, shaped in corresponding sets of rhymed or even unrhymed verses. It consists of bundles—to use the least flattering term—of lines—bundles quite arbitrary in size or number, but closely connected by assonance—that is to say, identity of vowel sound in the last syllable, but independent of the agreement in consonantal clothing which rhyme requires.

Now, the difficulty of competition under the first of these heads—that of language—rests upon all competitors in modernising or translating, and indeed is only an intense form of the general difficulty of translation itself. I do not propose to say much about it, though I think Captain Scott Moncrieff has wrestled well with it. It is the metrical and generally prosodic character which is so specially hard to retain. Translators have, naturally enough, tried all sorts of outflankings in their attack; but the worst point of these is that the adventure is not achieved, only evaded. If you do not convey the steady, fearless, ruthless tramp of the single line repeating itself; if you fail to reproduce the dropping fire of the assonance with its strangely combined advantage of repetition in the individual laisse or bundle, and freedom from monotony both in character and in quantity of sound in the several laisses—you do not give the effect of the Chanson de Roland to those who do know it, while you give something