‘History is the essence of innumerable biographies’, wrote Thomas Carlyle (1795–1881), while Schlegel (1772–1829) commented that ‘A historian is a prophet in reverse’. These are precisely the problems that beset any study of early music: much of the surviving music is anonymous, and because so little of it, comparatively speaking, has survived, the retrospective prophecies of music historians are based on fragmentary evidence. Consider the history of Spanish polyphonic song: no songbooks of Castilian-texted songs survive from before the end of the fifteenth century. Between them, these songbooks contain some five hundred songs; that is a fair number, one might think, but it is only the tip of the cancionero iceberg, to judge by the quantity of lyric verse dating from the period 1450–1530, the time when the first flowering of Spanish polyphonic song apparently occurred. However, these songs represent a notated repertory: there is plenty of extra-musical evidence to suggest that an improvised (or at least unwritten) song tradition existed much earlier in Spain. Any prophecies in reverse about developments in song forms, styles or themes must therefore take account of a practice we cannot hope to explore or describe.