In 000128 MMDigest Andy Taylor said of half-pedalling:

> Of course, pneumatic pianos can not do this.  This is just one
> example of the many reasons that player rolls, no matter how well
> done, will only approximate the "live" performance of a concert
> pianist.  There are many other reasons.

Obviously holding dampers half off the strings isn't possible -- doing
that needs feedback to the player's ears -- but a considerable number
of classical Duo-Art rolls certainly use flutter-pedalling.  Hofmann
is said to have insisted on it.

The story is that when he demonstrated it on the New York recording
piano, using full pedalling in stabs about a fifth of a second apart,
the playback gave too much pedal, so the editors learnt to shorten the
stabs to get the right effect. This gave them an insight into the
inaccuracies of recording normal pedalling, which were, as a result,
corrected.

However, the pedal pneumatic characteristic on Aeolian pianos was never
properly specified, so as Andy says, it was an approximation.  The
Buffalo Convention in 1908 obviously overlooked this !

A noticeable effect on Hupfeld pianos, when playing American or British
hand-played rolls, is for the pedal to stay on much longer.  The pedal
perforations also have to register properly.  On Hupfeld pianos,
probably to help cram in the Triphonola dynamic ports, the "sustain"
port is about 1.5 mm further to the right.

Dan Wilson, London