I've spent the last day proofreading a short viol piece by F. It's been an interesting experience, pursuing another composer's processes - constructing a piece according to another's blueprints.

Apart from developing some theories about accidentals that might bear repeating to some composition students some day, it struck me that there is a great difference between F's processes and my own.

When I 'process' some material (whatever that is - Satie, a pitch row, random numbers), any error will not produce anything very divergent from the overall effect. It has to be a big deal if I want to be exact because usually the level of process that minces the original material translates it to such an extent that it is largely unrecognisable (think of the solo piano opening of the second movement of Lovesongs, where left-over Wagner from the 1st movement becomes something resembling Berg - accidental but beautiful and fortuitous conjunction). F actually wants you to hear the lines, and follow them even though they've been scrunched into semitones and (in 2 voices) have been shuffled like cards.

Why do I use pre-existent material or even processes if I'm that indifferent to accuracy? I'd like to point out that in terms of accuracy, I don't mean that I'm indifferent to errors in the score: 20 notes in a 21:17 tuplet for example, or a botched metric modulation. I'm referring to mistakes that have been made in a transformation process or a number sequence mistakenly reconstituted. There is a margin of error that I'm willing to accept. It's almost an economy of effort. I check over what I've done to some degree, but I feel no compulsion to go over it to the degree that I am with F's piece. I remember finding a big mistake in the 'perfect' durational sequence in my 'opus 1', Supplication for mixed choir. It was in one part (the tenor part) and I only found it while we were preparing it for performance. I went home and tried to fix it but realised that if I did that, the entire framework was weighted in a completely different way (from that point on, the tenor line would be completely different) and would make my coda seem 'wrong'. In the end I concluded that I preferred it with the error in place. The 'sound' of the piece wouldn't be very different if the correct numbers had been used, but the change would alter the detail of the piece to such an extent that it would have to be recomposed. And I didn't want to do that. What would be the point of completely recomposing a piece so that it was 'correct' even though the soundworld remained the same, and would have a different weight (and therefore potentially a different 'feel') to the already composed version? I kept the error. Given that a major inspiration for the piece had been the Liturgie de cristal from Messiaen's Quatuor pour la Fin du Temps, the error is fortuitous. I was always slightly dubious about Messiaen's claims to the 'perfection' of the number 7 in this work, and the 'super-perfection' of the number 8. The error becomes the grit in the oyster, the human failing that makes the piece mine.

I'm not sure quite how I'll justify all of this in my viva but I'm working on it, which is the main thing!