The only excuse (besides mere unavoidable ignorance, which, no doubt, is a sufficient one) for this neglect is the curious fact, in itself adding to their interest, that these _chansons_, though a very important chapter in the histories both of poetry and of fiction, form one which is strangely marked off at both ends from all connection, save in point of subject, with literature precedent or subsequent. As to their own origin, the usual abundant, warm, and if it may be said without impertinence, rather futile controversies have prevailed. Practically speaking, we know nothing whatever about the matter. There used to be a theory that the Charlemagne Romances owed their origin more or less directly to the fabulous _Chronicle_ of Tilpin or Turpin, the warrior-Archbishop of Rheims. It has now been made tolerably certain that the Latin chronicle on the subject is not anterior even to our existing _Chanson de Roland_, and very probable that it is a good deal later. On the other hand, of actual historical basis we have next to nothing except the mere fact of the death of Roland ("Hruotlandus comes Britanniae") at the skirmish of Roncesvalles. There are, however, early mentions of certain _cantilenae_ or ballads; and it has been assumed by some scholars that the earliest _chansons_ were compounded out of precedent ballads of the kind. It is unnecessary to inform those who know something of general literary history, that this theory (that the corruption of the ballad is the generation of the epic) is not confined to the present subject, but is one of the favourite fighting-grounds of a certain school of critics. It has been applied to Homer, to _Beowulf_, to the Old and Middle German Romances, and it would be very odd indeed if it had not been applied to the _Chansons de geste_. But it may be said with some confidence that not one tittle of evidence has ever been produced for the existence of any such ballads containing the matter of any of the _chansons_ which do exist. The song of Roland which Taillefer sang at Hastings may have been such a ballad: it may have been part of the actual _chanson_; it may have been something quite different. But these "mays" are not evidence; and it cannot but be thought a real misfortune that, instead of confining themselves to an abundant and indeed inexhaustible subject, the proper literary study of what does exist, critics should persist in dealing with what certainly does not, and perhaps never did. On the general point it might be observed that there is rather more positive evidence for the breaking up of the epic into ballads than for the conglomeration of ballads into the epic. But on that point it is not necessary to take sides. The matter of real importance is, to lay it down distinctly that we _have_ nothing anterior to the earliest _chansons de geste_; and that we have not even any satisfactory reason for presuming that there ever was anything. [Sidenote: _Their metrical form._]