Interesting comments yesterday from both Don Cox and Craig Brougher.

When it comes to the 'correct' speed rolls ought to be transcribed
at, I'm not sure there's a definitive answer.  The problem comes when
modern-day ears try to mould an old performance into a new style --
I've noticed pianolists in the UK tend to homogenise the tempos of
pieces, irrespective of the marked speed.  What this guarantees is that
the original intentions will never be followed!  How many owners don't
even bother to calibrate their speed controls, setting tempo by ear?

This trend most particularly affects extremes of tempo.  I have a Duo-
Art roll (Hofmann playing Beethoven's 'Rage over a lost penny') that is
marked at 100 or so but has '30' pencilled on -- yet when Evgeny Kissin
played it in concert recently he was significantly faster than Hofmann
at marked speed.  The previous owner either had a dreadful tempo
control or (more likely) simply had no idea about what the music should
sound like!

When Craig Brougher discusses the compressed sound on the 78s, it's
certainly why the new recordings come out louder.  However, because
piano tone changes as dynamics increase, it's still possible to hear on
the 78 that the note was struck harder, even though it doesn't come out
much louder.

That's why it's so hard to work out just how true new recordings are,
compared to the original "what would have been heard in the studio on
the day".  The apparent fidelity achieved on the latest CD is all the
more remarkable when you consider that the Rachmaninoff Ampico rolls
actually had their dynamics created by Edgar Fairchild!

It would be nice if Wayne would let us know how he did it, now that
he's wreathed in triumph!  I suppose the real question is how much
fixing and manual correction was needed -- or could any of us scan
rolls and play them back in the same way without intervention, if
we could figure out the conversion algorithm?

Julian Dyer