A few points about Duo-Art regulation that I think should be stressed.

The knife valve bushings in the Duo-Art expression box -- both internal
and external -- must have no slop, otherwise the smallest increments of
the accordions will not move the knife valve.

The knife valve itself must seal its port fully at every possible
position: remember, the knife describes a complex motion across the
port under the influence of the accordion and the regulator movement.
If necessary, lap the knife and its orifice plate on very fine abrasive
paper on a sheet of plate glass until they seal perfectly, and
re-lubricate with graphite.

It's possible to alter the angle of the external knife valve cranks
with respect to the internal crank and the accordion crank.  Aeolian
surely had production settings for this.  If an overall acceleration in
the motion of the knife valve is achieved, where the movement is least
at powers 1-2 and greatest at 15-16, the dynamic curve will be slightly
concave, and the build up more subtly.  Beware though: over adjustment
and a weak pump will result in undesirable top end droop -- an S curve!
-- when the piano will certainly not meet the chord tests.  Use a
vacuum (water) gauge to get this right.

If the theme and accompaniment regulator springs are identical and have
the same initial tension, and the two dynamic curves are offset by the
zero adjustment, they will only parallel each other, not perfectly
interlace.  Then the theme will not be clearly differentiated at higher
dynamics and the chord tests will not be met.  To achieve proper
interlacing -- i.e., a given theme dynamic always halfway between
adjacent accompaniment dynamics -- the _rate_ of the theme spring must
be slightly greater than that of the accompaniment spring so the theme
curve 'pulls away' just sufficiently.

Note that the regulator spring adjustment design of the Duo-Art
expression box (unlike Welte and Ampico) means that tightening the
spring doesn't just raise its tension but by shortening it also
increases its rate.  British and early US boxes have identical theme
and accompaniment springs, so to achieve interlacing the theme spring
needs to be sufficiently tighter at rest than the accompaniment, and
its zero level adjusted appropriately.

Later US-design boxes seem to have had have the desired interlacing
pre-ordained (easier for technicians) by using a slightly higher rate
("stronger") theme spring.  These require the same initial tension but
the appropriate extra theme zero level adjustment.  Too tight a theme
spring makes the theme dynamics overtake the accompaniment's at the top
end.  Judicious, reiterative regulation is necessary!

Interestingly, unlike compression springs, to keep their coils together
all tension springs have an inherent initial tension which must be
overcome before the coils start to open.  But in well-made springs this
is minimal and the rate thereafter as prescribed.  Bear in mind that in
any replacement Duo-Art springs the inherent initial tension must be
less than that required to make the stack play at zero, or correct
adjustment will be impossible.

There is no case ever for reducing the overall 15/16" movement of the
accordions; the knife valve requires the full movement to produce the
correct dynamic curve.  It is the regulator spring rates that match the
dynamic curves to the dynamic range of the piano, by allowing the full
range of regulator movement for a chosen maximum pump tension, which
determines the piano's fortissimo.

Only a small amount of regulator spring rate increase is possible in
the Duo-Art expression box design.  Overtightening risks the spring
exceeding its elastic limit at power 15 and becoming permanently
overstretched, so different rate springs of the same length were used
for different sizes of piano.  There seem to have been perhaps 5
different 'strength' regulator springs stocked by Aeolian in England.

Many English Duo-Arts have a suction tension regulator ("unloader")
immediately after the pump which must be set sufficiently high.  Note
that insufficient tension prevents true fortissimos and actually
impairs pianissimo performance.

The full dynamic range of the piano may be defined as from a true
pianissimo, without notes missing, to a fortissimo just less than when
the (correctly voiced) piano starts to "jangle".  A limited Duo-Art
dynamic range may appease the non-musical but is an injustice to the
original pianists' performances and the whole notion of a reproducing
piano!

Many instruments seem too loud because their hammers have hardened with
use and they have not been correctly re-voiced.  Please note that while
allowing ravishingly soft playing, correct voicing does not prevent
powerful fortissimos.  It actually enhances dynamic contrast, making
the piano seem less strident while allowing clear articulation.  Would
that most modern -- even high quality -- pianos were adequately voiced!

Patrick Handscombe
Wivenhoe, Essex, UK