Apropos Pat DeWitt's proposal for configuring the "quieting" features
of Duo-Arts, further commented upon by Ralph Nielsen, I have the
following observations:

In my rebuilding career I have encountered only one example of
a Duo-Art installation with both hammer rail lift and keybed shifter
pneumatics.  It was in a Steinway of unremembered but probably 'OR'
length, and unremembered but probably early 1920s vintage.  As far
as I could tell, the factory tubing configuration was in place when
I started the restoration, and it had the keyshift pneumatic tubed
to hole #95 and the hammer rail lift pneumatic tubed to the keyslip
modify switch.  Being unable to find any references (this was mid
1970s) that were definitive, I decided to reverse that configuration
for the following reasons:

1. The vast majority of Duo-Art installations had _only_ the hammer
rail lift.  It therefore seems logical that the coding editors would
have coded for that configuration.  And as mentioned previously, the
hammer rail lift is more responsive and power conserving than the
keybed shift.

2. Perhaps a more important reason comes from this analysis: When the
hammer rail lifts, there is lost motion (or a gap) introduced between
the key pneumatic pushrods and the underside of the key tails, and, not
insignificantly, the geometry of the hammer action is affected.  This
causes the pushrods to impact the key felt with more force than when
the hammer rail is down and the lost motion is (or should be)
negligible.

If, as frequently happens, the modify switch is then left in the soft
position for long periods of time, either through forgetfulness or
preference, accelerated wearing of both the key felts and the action
parts will result, leading to regulation problems which can be both
expensive and time-consuming to correct.  (Ampico cleverly minimized
this problem in the "B" stack by having the "soft pedal compensation"
pneumatics which raised the resting level of the key pushrods to stay
close to the underside of the keys.  The keys still visually dip when
the hammer rail lifts, but there is no resulting lost motion.)

3. Since no lost motion is introduced when the keybed shift engages,
I also eliminated having the #2 Accompaniment activate from the modify
switch.  This allows the two "softening" effects of the keybed shift
to operate at best advantage, i.e., hitting only one or two strings
instead of all, and hitting those with a different part of the hammer
head felt which is presumably less compressed and therefore softer than
the unshifted area, and unaffected by the arbitrary increase in power
(#2 Accomp.)  On a hammer rail only installation, that linking is still
needed because the lost motion gap has the effect of reducing the power
response curve by approximately that amount.

John Grant