While the Duo-Art system was conceived by Votey of Aeolian in the US,
and the first English-made instruments were fitted with American-made
expression boxes, soon after WW1 the largely independent British
Aeolian Co engineered and manufactured its own versions.

Performance differences experienced between the new British designs and
post-war US boxes were the subject of good deal of correspondence
between editors Creary Woods in New York and Reginald Reynolds in
London.  This correspondence and physical comparisons of the different
boxes reveal much about their intended operation and characteristics.

It became clear that the British engineers and recording editor had
adhered to the original principles and characteristics, ignorant of the
fact that, for reasons which seem hard to understand, changes had been
introduced in the US, such as the grand crash valve.  Creary Woods, who
had helped develop the Duo-Art, expressed disapproval of this feature,
which is superfluous and absent from early US and all British designs,
explaining that it had been tried on some concert instruments and
unfortunately applied to production models.  Woods urged the British
to eschew it.

All British and early US expression boxes have identical theme and
accompaniment regulator springs.  Thus if both springs have the
same rate and initial tension and the knife valves are set the same,
the theme and accompaniment expression curves will be coincidental
throughout.  Increasing only the theme knife valve setting will raise
its whole curve (providing that the pump tension and flow are
sufficient) by a more or less 'parallel' amount throughout the dynamic
range.

Each theme suction level is then greater than its corresponding
accompaniment level, but less than the next accompaniment level above,
which maintains the interlace logic of the original concept that no
levels are duplicated (to provide the maximum number of gradations) or
redundant (since the theme level can never fall below that of the
accompaniment because of the flap valves).

However, the human ear is more sensitive to changes in low sound
pressure levels than in high, and a slightly concave dynamic curve is
desirable, as Ampico stressed, to provide more gradations and thus
subtlety at low dynamics.  Such curves are indeed given by British and
early US expression boxes when their knife valve and spill linkages are
optimally adjusted.

Then to ensure that the theme interlaces distinctly at the higher
dynamic levels its curve must be steepened by judicious, minimal
tightening of its regulator spring (and appropriate reduction of the
knife valve minimum setting).  This adjustment seems to have eluded the
service technicians in the field with the result that the theme often
became indistinct in louder passages, and later US boxes were fitted
with different rate springs to make the theme stand out automatically.
Those responsible seem often to have exaggerated this remedy so that
even with careful adjustment the theme still pulls strongly away at
higher dynamic levels, and performances, even with later rolls, can
verge on being strident.

Stack suction readings reveal that a musically convincing British
Duo-Art has every theme level rather nearer the next accompaniment
level above than that below.  It is not possible to give absolute water
gauge readings as a number of different size stack and expression box
combinations are encountered and action regulation and weight of touch
vary.  For comparison, one very early 1914 US Steinway O of which I
know plays fortissimo as loud as a live pianist at 16", while a 1925
Hamburg O with a smaller stack requires a massive 25".  Corresponding
pianissimo readings are 2.5" and 3".

Both pianos have original and different weight hammers and key leading,
are exquisitely regulated, and have graduated stacks whose pouches are
super-sensitive and valves have just sufficient travel to allow
unimpeded repetition.  Most importantly, they are voiced not like
modern Steinways but as nearly as models of their periods were (from
ear-witness testimony and gramophone recording comparisons), so that
they have an additional inherently steepening dynamic characteristic:
they have a subtle and velvety pianissimo and a creamy and articulate
mezzoforte, but "the tone comes up" progressively and effortlessly
towards fortissimo.

Understanding the natural dynamic range of an instrument which would
be utilised by a live pianist and making the Duo-Art expression system
match it is crucial to obtaining convincing, musical reproduction.
It should not be forgotten that many instruments were designed to play
in large, heavily draped rooms which soaked up sound.

All except the earliest British Duo-Arts have a regulator/spill in the
suction supply line whether they have a rotary (positive displacement)
or centrifugal pump, and seem the better for it.  Regulating the
expression box spill correctly is essential to ensure the proper
build-up of the curves and the vital modulation of the accompaniment
levels by the theme below power 10.  (The Duo-Art is actually the
simplest yet most sophisticated of all the reproducing systems.)

Using a jig and gauge I have never experienced any difficulty adjusting
the accordions to move exactly in 1/16" steps.  Needless to say, all
box bushings should be free but with no slop.  If all is correct the
play/won't play chord tests on the appropriate test roll will usually
be met successfully.  And somewhat amazingly, a British Duo-Art with
its theme set slightly on the high side, but still interlacing
throughout, will play US rolls convincingly without the theme
disappearing!

Patrick Handscombe
Wivenhoe, Essex, UK