TUNES 43 from the gravest musicianly error, which is to set up a standard tune for a ballad, incorporating variants and ignoring different versions. The heresy of the standard text vitiates both words and music of the otherwise admirable Romancero of Doncieux and Tiersot. To pursue this theme of the difficulty of transcribing tunes accurately, we must further notice that the tunes are never quite constant, either throughout the individual ballad or at different performances. If the air has been fixed on paper, we still do not know its subtle application to the ballad. To meet this difficulty recording is now done by phonograph, and the records are then stored and interpreted at leisure. Even this process appears to be insufficiently delicate to capture all the nuances of Balkan singing, with difficult rhythms, irrational intervals, optional flourishes, and other complexities. C. Obreschkoff demands for this purpose the preparation of sound-films. Among the advantages that might accrue from so elaborate a process might be some certainty as to whether the intervals of one-third and quarter tones are really musical intervals or spoken. They occur in passages so rapid that the ear cannot precisely determine their value. In western coun- tries this refinement does not occur, and the greatest difficulty appears to lie in the recognizing of modes and of the pentatonic and heptatonic scales. It is therefore extremely difficult to describe ballad tunes; how much more to compare them. I cannot do better than repeat Tiersot's statement (op. cit, p. xliii); he is speaking of the typical melody of his ballads. It is indeed usually very difficult to recognize this type. There are some melodies so distantly related that one would not expect to find anything in common in them, were they not associated with the same words. The tonality is incessantly changed by the varying sentiments of the popular singers. There are melodies known to us in an equal number of major and minor fragments: how embarrassed we are when we must say which of the two modes is to be adopted! Then there are alterations modifying the ancient scales, or introducing, on the other hand, into modern melodies intonations contrary to their spirit. Even koff) wrong rhythms: 3/8 for 7/16 and 5/8 for 9/16. The error would be impor- tant for an account of Bulgarian music, but not for international comparisons, since rhythms are very variable, without affecting the melodic curve. Of another collector who has done yeoman service in the matter, it was complained that his transcriptions were uninteresting.