I agree with Doug Henderson's assessment of a Duo-Art mechanism in a
Steinway DR concert grand in principle.  The D/A mechanism was basically
designed for parlor performances of the classics.  The rolls were
likewise coded as well.  The pump was designed to slip its belt at high
pressures, acting exactly like a pressure regulator to prevent its motor
from stalling.  Everything we are talking about here is easily
measurable.  This is not empty speculation.  For example, were you to
prevent the belt from slipping, and ran the pump a little faster for a
larger piano, you would eventually burn up the motor, and probably start
a fire.  You have to have a special mechanism to play a piano which has
twice the musical power capacity of a baby grand.

If anybody is interested in the physical measurements as they apply to
the mechanism in a dynamic situation, they should measure (as I have) the
pressures at full crash with 3 notes being played, and then with two
handfuls of huge chords, being played one right after the other.  The
pressure drops to less than half the pressure of the individual crash
(ffff) notes momentarily-- at the instant they are being played.  You can
also calculate it by displacement/time.

To play a _concert grand_ realistically at the concert stage level, you
would have to maintain about 80 inches of water for any large group of
chords or barrage you decided to attack that keyboard with.  Two pieces
come to mind which are decidedly different, and which both use crash
intensities, are Danse Rituelle Du Feu (Ritual Fire Dance) and Bolero.
The DR should have no problem playing the first piece to its full
potential (at least from a physical perspective) because its loud, full
power notes are individually played.  The second piece it could not
possibly attempt, because no Duo-Art pump is capable, without modifi-
cations, of performing this massive, atonal work.

Much of the music played by Hofmann, therefore, cannot be realistically
performed without someone also "faking it" on the levers while it is
being played (or, as I said before, a modification of the coding on the
roll).  Even then, the pump will have to be speeded up, and a few other
modifications made, such as a real, genuine spill valve safety to prevent
stalling the motor, instead of the slip-shod slipping belt thing that was
good enough for home use, but could never used for a realistic
professional performance on the concert stage.

As far as hand manipulation of levers for expression is concerned, I
fully appreciate what can be done in that department.  I have heard some
very fine pianistic effects, and it is highly realistic.  However, few of
us are able to do it that way.  It takes years of practice.  Also, if you
were to remove the return spring tension to those levers and snug the
nuts so they moved freely with the accordions as the roll operated them,
you would sit then at the bench and wonder how in the world any human
hand could move and precisely position those levers that rapidly.  In
some music, those accordions are moving very quickly.  So it gives you a
new appreciation for interpretive and arranged coding, in that the ear,
which accepts these nuances as genuine, is easily fooled as well.

Craig Brougher