First, a correction to my last posting on MMD 970820.  My plan for a
punch included a stepper motor with an advance of 0.001 _thousandths_ of
an inch, not 0.001 _percent_.  0.001 percent would be a bit of overkill.

Re Wayne Stahnke's comment about the discussion getting too technical for
many readers.

Feedback from MMD'ers would help here.  I think about how I read the
Scientific American, for instance (or for that matter, the MMD!).  I
don't read every article in detail, only those in which I'm interested.
Others I may scan, or skip altogether.  We don't know who may be
interested in what, so I'm inclined to present more than less.

Re Paul Johnson's comments on what inaccuracies can be heard by the
average listener.

I'm reminded of the time when I was printing the AMICA bulletin on a
really minimal offset printing press in our kitchen.  I was quite
obsessive about pulling out every page that had the least defect.  One
day I was reading the newspaper and realized I had read a page with a
large, weakly printed area, and had never noticed it.  I simply read for
the information and ignored the defect.

Now enlightened, I let minor defects in the bulletin pass, and found
after inquiries that _no-one_ noticed.  As several people have noted, we
can, and usually do, enjoy imperfect things.  It's only after a certain
threshold is reached that we notice the defects.

 [ Printed matter -- the newspaper -- is a type of quantized data:
 [ there are only 26 letters.  If the ink is faint but still legible,
 [ then the reader nonetheless can determine what the original data
 [ was.  A better analogy might be a newspaper reproduction of a
 [ photograph.  If the ink is faint it can be quite difficult to
 [ determine the original data.  -- Robbie

We strive for perfection for personal reasons.  Speaking only for myself,
I want to come as close to the original intent of the makers of
mechanical music as I can, without getting totally crazy about it.  And
only because it pleases me to do so.  I will spend hours with Photoshop,
for instance, restoring an old half-tone picture down to the dot level.
No-one but me will ever appreciate the result, but it pleases me.

Can the inaccuracies of roll copying be heard by the average listener?  I
think so.  If he is focused on the music, and not simply hearing
background music, he can usually hear a difference between an accurate
copy and a less good one.  Maybe he can't tell exactly what is different,
but he can tell that there is a difference.

Re Joerg Wendel's comments on roll dimensions.

It is certainly true that the dimensions of original paper roll copies
cannot be trusted.  That is why we need to recreate the masters as well
as we are able.  There is sufficient information on most rolls to deduce
what the master probably was.  For example, scallops on the note holes
and roll tempo markings can tell us what was the punch advance.

Joerg's point is well taken that to recreate the original intent we must
have a piano that is identical to the recording piano.  This means make,
size, regulation, voicing, etc.  The artist adjusts his playing to the
piano and environment, which means we can never exactly recreate the
original intent.  The closest we can come is to recreate the master roll.

( -Sigh-  Perfection eludes us again. )

Bob Billings