The tale of Paul Dukas's cripplingly critical view of his own music is familiar to scholars: when his friend George Enescu enquired in early 1935 as to the whereabouts of his unpublished manuscripts, he reportedly replied that he had burned them all.' The truth of that story is confirmed by the fact that the sole unedited Dukas scores in existence today are a handful of student efforts dating from his time at the Paris Conservatoire. When coupled with the reality that the composer only published 13 works, the avenues available to researchers seem frustratingly restricted. An important but little-known detail, however, is that Dukas's dissatisfaction with his works did not encompass all aspects of their conception. A sizeable number of scenarios, outlines. scenery sketches - and even a whole, typed libretto for proposed works, which were often at least partially developed before being obliterated, are still intact today. These rich resources remain underexploited. Consequently, there are gaps in Dukas scholarship, notably on the subject of his later years where the paucity of finished scores is most striking.