When I first started making MIDI files from 176-note Duo-Art organ
rolls, I was forced into using a pneumatic roll reader rather than an
optical reader.  By "forced" I mean the width of optical scanners in
existence then were too narrow for the 15.25 inch width of my rolls.

Realizing that pneumatic roll reading was a mechanical process, the
possibility for error lead me to classify my scans as less than perfect
for recreating a "master" as some of the optical scanners now do.
Still, my process was good enough to equal or exceed the results of
just playing a roll the old fashioned way and that was my purpose to
begin with.  I use these files to play my Duo-Art organ, and the paper
rolls stay on the shelf.

Recently I started the development of a process to verify the accuracy
of my scanned rolls.  The study of the duration of a single dot
perforation in my process, compared to the theoretical, revealed that
my roll scans do indeed contain enough information to be of high
accuracy archival quality.

There are two characteristics of the 176-note roll that make it unique.
First, the tracker bar has two parallel rows of holes that are exactly
0.25 inches apart, on center.  These two rows correspond to the upper
keyboard and lower keyboard of the organ.  The factory perforator had
to be set so that the step rate could produce holes simultaneously,
if needed, in both rows.  That means the step rate had to be divisible
by four.  Thus, the step rate of the rolls, which is 28 steps per inch,
gives exactly 7 steps between the rows.  All Duo-Art rolls use this
same steps per inch.

The second unique characteristic is the type of perforation used for
the Duo-Art coding.  The coding holes for the swell shades are always
the same length, and thus are always made with the same number of
perforation steps.  These coding holes occur frequently throughout the
entire roll.

What this means is that these rolls carry a "Rosetta Stone" in the
coding.  Using the length of the roll (the time it has been playing)
and accounting for take up spool build up, if the tempo has been
correctly set, the exact duration (time) of every swell shade code
hole can be calculated.  With enough calculations, the accuracy of the
pneumatically scanned roll can be verified and corrected to recreate
a factory master of close tolerance.  If necessary, the roll length
can be actually measured and that length used in the calculations.

If the need arises, a computer program can be written to produce
a master roll from my existing pneumatic roll scans.  This is a bonus
I never expected.

Bob Taylor