Fractal flourishes end in musical discord

John Barrow suggests that musical compositions, when viewed from the appropriate perspective, embody 1/f noise characteristics (THES, June 25), a conclusion drawn from data given by Voss and Clarke in a 1975 paper. There are some workers in this field who are not wholly convinced by Voss and Clarke's choice of parameters as musical correlates, or by their findings. Indeed, one hard-bitten group of industrial-research chemists I lectured to about fractals and music cynically suggested that 1/f log-log plots can all too easily be drawn from dubious data!

More disturbing, however, is the use to which this evidence is often put. Voss and Clarke find a number of "modern" works by composers such as Jolet, Stockhausen and Carter which do not evince 1/f characteristics. Are we to regard such renegade works as being somehow contrary to nature? Are works which do show a 1/f tendency (and are therefore, to use Professor Barrow's term, "appealing") somehow validated by their supposed affinity with nature?

Similar naturalist theses (such as the golden section and Fibonacci series) regularly invade musicology, and can be misused by less than fully numerate theorists in their attempt to "scientifically" justify musical works or genres. While 1/f noise and other fractal-based methods often provide useful algorithmic sources for computer-generated music, it may be that they are no more helpful to the understanding of the musical experience than is the appeal to the "divinely inspired improvising genius".

David Cooper

Lecturer in music

University of Leeds

d.g.cooper@leeds.ac.uk