A very interesting observation by Paddy Handscombe involved the
differences in English Duo-Arts from American models and how they
also vary in dynamic range.  I do not doubt this a bit, and when the
crash valve is not used, it is clear to me that a powerful ffff is
not available or desired; but musically, I do not say that it isn't
used and was never a part of the classical repertoire.  It was.

So I want to return somewhat to the original design model of the
American Duo-Art and explain something very basic about it.  It's
Theme and Accompaniment curves are designed ideally to be straight
lines.  Forget the fact that you can make them curve, even those that
don't want to (and some boxes display an almost straight line without
any easing and spring adjustments.  That ought to say something).

The reason both boxes have large-diameter long springs is so that at
no time will the regulators overstretch them (when they are correctly
tightened, that is.  I am not referring to these boxes whose spring
adjusters have already subtracted an inch of the spring length just to
get the piano loud enough not to miss any notes).  That means, the
designers obviously intended the regulator to have a linear rate with
respect to its displacement at whatever position the intensity and
number of notes being played called for.  I know Paddy agrees with
this, because he said:

> 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.

If Paddy means by this that both springs were always equal in rate,
then I heartily disagree for pianos destined to the American public.
The only ones I have ever seen like that were small baby grands which
were built late, and very few of them, actually.  I doubt that I have
come across more than a dozen, in all the 30-some years of rebuilding.
But his point that whatever the spring rate happens to be, is also the
rate of the box, is correct.  Therefore, were the springs to be equal,
there would be no other way to favor theme, unless you were to
transpose the theme curve high enough above the accompaniment to make
a difference, and then they would be parallel, separated only by a
difference in zero intensity, which cannot be over 1/2".  That is
incorrect, and also does not allow enough difference at high intensities
for the piano to favor theme over accompaniment since everything is on
a percentage basis anyway, and what's a half inch, more or less, to 25
inches?  The piano isn't that accurate, so what you have, in effect, is
a Themodist.

In other words, The rate of a charted line is its tangent angle
subtended to the horizontal.  Were the spring being overstretched
during the regulator's travel, the line would curve upward increasingly,
and then almost top out before reaching its zenith.  That becomes the
lazy "S" curve I spoke of earlier.  That same characteristic curve is
also called hysteresis, and one characteristic is a relatively slow
takeoff from the zero intensity setting at the lower left.  So with a
hysteresis curve, you have all the elements you do not want to
encourage, and just the opposite of what the ear would acknowledge as
genuine, to boot.  It would not follow any portion of the log curve
that I am familiar with-- which Paddy alluded to.

A linear Duo-Art does _not_ prevent more subtle changes at the lower
volumes -- just the opposite, in fact.  The sensitivity of the ear to
subtleties at low intensities is inherent to the ear's circuitry and
does not require an engineer to design a lazy S curve to accommodate
it in his expression box.  If you will study the dB (intensity) curve
versus vacuum pressure, you will notice an amazing fact-- a linear
vacuum pressure curve at those low to medium intensities produces
something close to the exponential dB curve (after a fashion).  However,
as you go up the scale, the relationship goes bye-bye.  Which means (to
me) that it's a matter of arranging and also average overall power
anyway -- not perfect dB adherence.

It really doesn't matter if a dB curve follows the box's characteristics
anyway, since the arranging is all done on one's own sliding scale and
that doesn't have to be any particular log curve.  And now, here's the
clincher-- the dB scale (sound intensity) is a moot point!

All reproducers ignore the strict dB intensity scale, as far as their
dynamic is concerned, and most especially the Duo-Art.  How can I say
that?  Because decibel is just a measurement of a peak intensity, but
overall, the volume of the piano (that's what concert goers respond to)
is the integral of the intensity with respect to the time it stays
loud, plus the addition of other aspects, like a wider range of other
frequencies all playing at once (in other words, average power).

Loudness depends on how many notes are playing at any given time for
how long, and herein lies the ability of the piano to duplicate an
actual performance.  While the dB of any given note may be the same,
or actually dropping, due to pump demands, large chords are "louder"
than single notes to the ear.  So forget dB (which is basically
impulse-related).  The linearity of the box is there to accommodate
the largest number of notes the piano can play at any given time at a
given step and return instantly to the zero intensity without wide
differences in time lapse (as you would cause if you scaled the box
characteristic to curve).

So this is how we know it should not curve.  It must either be flat, or
its return timing varies, depending on its physical positioning at the
time (since zero intensity regardless of the number of notes to be
dropped or played is always the same, but the position of the regulator
is not).  And then when you look at the springs, you see just that.
They cannot be overstretched (since they are 3" long unstretched and
can stretch to 6" before they become non-linear), so that means their
rates are linear-- simply by a plain physical principle that springs
are linear when stretched lightly.  A light stretch means they do not
deflect more between coils than the diameter of their own wire.  Sorry,
but that's just a fact.  The rate of the spring determines the rate of
each side of the box, respectively.  To that, you must agree.  So if
the rate of that spring is flat and linear, then the rate of the box is
flat and linear, or, something is not right.

Regarding the missing crash valve on British Duo-Arts: we here in
America understand your situation.  I would have agreed with Creary
Woods myself in case I wanted to sell lots of Duo-Arts in England, to
do away with the crash valve in British models because of the housing
situation common there, and my proximity to the neighbors.  Moderately
soft is a good idea.  And too, a concert pianist virtuoso will not tear
up anybody's piano under any conditions.  He plays to the room.  He
isn't out to tear up your instrument, but to play well in whatever
instance he finds himself in.  So when he is playing a small room, he
scales down the performance.  Besides, baby grands top out long before
concert grands do, so to pound loudly will not produce more tone or
dynamic once that piano reaches its limit.  (Think of pianos as an
amplifier.  An amplifier with more watts takes the same input and puts
out more power at the top end, even though the softest notes are about
the same).

As to sound intensity measurements made by recording studios, pianos
vary among themselves tremendously.  The loudest measurement I have
yet read about is an 1885 "Gottschalk" Chickering concert grand,
registering 126 dB (according to one record jacket).  The artist said
that his ears were still ringing 3 days later.  And as an additional
warning as far as sound pressures are concerned, the dB meter does not
necessarily measure pure dB, but the overall sound power of an impulse,
and as such are not necessarily germane to the audience's reaction of
loudness.  It's an "iffy" measurement that is directly proportional to
loudness, but that is all.

Craig Brougher