Now, even in ages of writing, it has been found difficult or impossible to keep linguistic novelties and novelties of metre out of old epics. We later refer (_Archaeology of the Epic_) to the _Chancun de Willame_, of which an unknown benefactor printed two hundred copies in 1903. Mr. Raymond Weeks, in _Romania_, describes _Willame_ as taking a place beside the _Chanson de Roland_ in the earliest rank of _Chansons de Geste_. If the text can be entirely restored, the poem will appear as "the most primitive" of French epics of the eleventh and twelfth centuries. But it has passed from copy to copy in the course of generations. The methods of versification change, and, after line 2647, "there are traces of change in the language. The word _co_, followed by a vowel, hitherto frequent, never again reappears. The vowel _i_, of _li_, nominative masculine of the article" (_li Reis_, "the king"), "never occurs in the text after line 2647. Up to that point it is elided or not at pleasure.... There is a progressive tendency towards hiatus. After line 1980 the system of assonance changes. _An_ and en have been kept distinct hitherto; this ceases to be the case." [Footnote: _Romania_, xxxiv. pp. 240-246.] The poem is also notable, like the _Iliad_, for textual repetition of passages, but that is common to all early poetry, which many Homeric critics appear not to understand. In this example we see how apt novelties in grammar and metre are to steal into even written copies of epics, composed in and handed down through uncritical ages; and we are confirmed in the opinion that the relatively pure and orthodox grammar and metre of the twenty Books must have been preserved by written texts carefully 'executed. The other four Books, if equally old, were less fortunate. Their grammar and metre, we learn, belong to a later stratum of language. These opinions of grammarians are not compatible with the hypothesis that _all_ of the _Iliad_, even the "earliest" parts, are loaded with interpolations, forced in at different places and in any age from 1000 B.C. to 540 B.C.; for if that theory were true, the whole of the _Iliad_ would equally be infected with the later Odyssean grammar. According to Mr. Monro and Sir Richard Jebb, it is not.