selection of MSS. or the like, has continued. Nevertheless, the number of chansons actually available is so large that no general characteristic is likely to have escaped notice; while from the accounts of the remaining MSS., it would not appear that any of those unprinted can rank with the very best of those already known. Among these very best I should rank in alphabetical order-Aliscans, Amis et Amiles, Antioche, Baudouin de Seboure (though in a mixed kind), Berte aus grans Piés, Fierabras, Garin le Loherain, Gérard de Roussillon, Huon de Bordeaux, Ogier de Danemarche, Raoul de Cambrai, Roland, and the Voyage de Charlemagne à Constantinoble. The almost solitary eminence assigned by some critics to Roland is not, I think, justified, and comes chiefly from their not being acquainted with many others; though the poem has undoubtedly the merit of being the oldest, and perhaps that of presenting the chanson spirit in its best and most unadulterated, as well as the chanson form at its simplest, sharpest, and first state. Nor is there anywhere a finer passage than the death of Roland, though there are many not less fine.

It may, however, seem proper, if not even positively indispensable, to give some more general particulars about these chansons before analysing specimens or giving arguments of one or more; for they are full of curiosities.

In the first place, it will be noticed by careful readers of the list above given, that these composiLanguage. Oc and oïl. tions are not limited to French proper or to the langue d'oil, though infinitely the greater part of

them are in that tongue. Indeed, for some time after attention had been drawn to them, and before their actual natures and contents had been thoroughly examined, there was a theory that they were Provençal in origin. This, though it was chiefly due to the fact that Raynouard, Fauriel, and other early students of old French had a strong southern leaning, had some other excuses. It is a fact that Provençal was earlier in its development than French; and whether by irregular tradition of this fact, or owing to ignorance, or from anti-French prejudice (which, however, would not apply in France itself), the part of the langue d'oc in the early literature of Europe was for centuries largely overvalued. Then came the usual reaction, and some fifty years ago or so one of the most capable of literary students declared roundly that the Provençal epic had "le défaut d'être perdu." That is not quite true. There is, as noted above, a Provençal Fierabras, though it is beyond doubt an adaptation of the French; Betonnet d'Hanstone or Beton et Daurel only exists in Provençal, though there is again no doubt of its being borrowed; and, lastly, the oldest existing, and probably the original, form of Gérard de Roussillon, Giratz de Rossilho, is, as its title implies, Provençal, though it is in a dialect more approaching to the langue d'oïl than any form of oc, and even presents the curious peculiarity of existing in two forms, one leaning to Provençal, the other to French. But these very facts, though they show the statement that "the Provençal epic is lost" to be excessive, yet go almost farther than a total deficiency in proving that the chanson de geste was not originally

Provençal. Had it been otherwise, there can be no possible reason why a bare three per cent of the existing examples should be in the southern tongue, while two of these are evidently translations, and the third was as evidently written on the very northern borders of the "Limousin" district.

The next fact-one almost more interesting, inasmuch as it bears on that community of Romance tongues of which we have evidence in Italian. Dante,1 and perhaps also makes for the antiquity of the Charlemagne story in its primitive form-is the existence of chansons in Italian, and, it may be added, in a most curious bastard speech which is neither French, nor Provençal, nor Italian, but French Italicised in part.2 The substance, moreover, of the Charlemagne stories was very early naturalised in Italy in the form of a sort of abstract or compilation called the Reali di Francia, which in various forms maintained popularity through medieval and early modern times, and undoubtedly exercised much influence on the great Italian poets of the Renaissance. They were also diffused throughout Europe, the Carlamagnus Saga in Iceland marking their farthest actual as well as possible limit, though they never in Germany attained anything like the popularity of the Arthurian legend, and though the Spaniards, patriotically resenting the frequent forays into Spain to which

1 V. the famous and all-important ninth chapter of the first book of the De Vulgari Eloquio.

2 See especially Macaire, ed. Guessard, Paris, 1860.

3 So also the geste of Montglane became the Nerbonesi.

Diffusion of the

the chansons bear witness, and availing themselves of the confession of disaster at Roncesvalles, chansons. set up a counter-story in which Roland is personally worsted by Bernardo del Carpio, and the quarrels of the paynims are taken up by Spain herself. In England the imitations, though fairly numerous, are rather late. They have been completely edited for the Early English Text Society, and consist (for Bevis of Hampton has little relation with its chanson namesake save the name) of Sir Ferumbras (Fierabras), The Siege of Milan, Sir Otuel (two forms), the Life of Charles the Great, The Soudone of Babylone, Huon of Bordeaux, and The Four Sons of Aymon, besides a very curious semi- original entitled Rauf Coilzear (Collier), in which the well-known romancedonnée of the king visiting some obscure person is applied to Charlemagne. Of these, one, the version of Huon of Bordeaux,1 is literature of no mean kind; but this is because it was executed by Lord Berners, long after our present period. Also, being of that date, it represents the latest French form of the story, which was a very popular one, and incorporated very large borrowings from other sources (the loadstone rock, the punishment of Cain, and so forth) which are foreign to the subject and substance of the chansons proper.

Their author

Very great pains have been spent on the ship and publi- question of the authorship, publication, or performance of these compositions. As is the case with so much medieval work, the great mass

cation.

1 Ed. S. Lee, London, 1883-86.

of them is entirely anonymous. A line which concludes, or rather supplements, Roland

"Ci falt la geste que Turoldus declinet "

has been the occasion of the shedding of a very great deal of ink. The enthusiastic inquisitiveness of some has ferreted about in all directions for Turolds, Thorolds, or Therouldes, in the eleventh century, and discovering them even among the companions of the Conqueror himself, has started the question whether Taillefer was or was not violating the copyright of his comrade at Hastings. The fact is, however, that the best authorities are very much at sea as to the meaning of declinet, which, though it must signify "go over," "tell like a bead - roll," in some way or other, might be susceptible of application to authorship, recitation, or even copying. In some other cases, however, we have more positive testimony, though they are in a great minority. Graindor of Douai refashioned the work of Richard the Pilgrim, an actual partaker of the first Crusade, into the present Antioche, Jérusalem, and perhaps Les Chétifs. Either Richard or Graindor must have been one of the very best poets of the whole cycle. Jehan de Flagy wrote the spirited Garin le Loherain; and Jehan Bodel of Arras Les Saisnes. Adenès le Roi, a trouvère, of whose actual position in the world we know a little, wrote or refashioned three or four chansons of the thirteenth century, including Berte aus grans Piés, and one of the forms of part of Ogier. Other names -Bertrand of Bar sur Aube, Pierre de Rieu, Gerard d'Amiens, Raimbert de Paris,