In sober truth, this Arab myth is but a bad dream of yesterday, a nightmare following upon an undigested perusal of the _Thousand and One Nights_. Thanks to Galland, Cardonne, and Herbelot, the notion became general that the Arabs were the great creative force of fiction. To father Spanish _romances_ and Provençal _trobas_ upon them is a mere freak of fancy. The tacit basis of this theory is that the Spaniards took a rare interest in the intellectual side of Arab life; but the assumption is not justified by evidence. Save in a casual passage, as that in the _Crónica General_ on the capture of Valencia, the Castilian historians steadily ignore their Arab rivals. On the other hand, there is a class of _romances fronterizos_ (border ballads), such as that on the loss of Alhama, which is based on Arabic legends; and at least one such ballad, that of Abenamar, may be the work of a Spanish-speaking Moor. But these are isolated cases, are exceptional solely as regards the source of the subject, and nowise differ in form from the two thousand other ballads of the _Romanceros_. To find a case of real imitation we must pass to the fifteenth century, when that learned lyrist, the Marqués de Santillana, deliberately experiments in the measures of an Arab _zajal_, a performance matched by a surviving fragment due to an anonymous poet in the _Cancionero de Linares_. These are metrical audacities, resembling the revival of French _ballades_ and _rondeaux_ by artificers like Mr. Dobson, Mr. Gosse, and Mr. Henley in our own day. On the strength of two unique modern examples in the history of Castilian verse, it would be unjustifiable to believe, in the teeth of all other evidence, that simple strollers intuitively assimilated rhythms whose intricacy bewilders the best experts. This is not to say that Arabic popular poetry had no influence on such popular Spanish verse as the _coplas_, of which some are apparently but translations of Arabic songs. That is an entirely different thesis; for we are concerned here with literature to which the halting _coplas_ can scarcely be said to belong.