Re: Offset of Perforations in Port Positions 0-3 in Artrio-Angelus Rolls

It is well known that ports 0-3 of the the Artrio-Angelus tracker bar
(the four leftmost ports) are offset with respect to the positions 
they would occupy if they were on 9-per-inch centers.  I measured the
offsets of these four ports in 2003 and I found that they are offset to
the left by 0.042 inches.  I posted a note about that result a few days
ago here on MMD [170528 MMDigest].

It is natural to expect that the rolls intended for use with these
tracker bars would have the identical offset for the ports in question,
which would ensure that the rolls would match the tracker bars exactly.
Today I did a brief investigation of two rolls, one early and one late,
to confirm that expectation.  To my surprise, the offsets in the rolls
do not match the tracker bar offsets.  Following is a brief report of
my findings.

For Artrio-Angelus roll 7510 ("Two Larks" by Leschetizky, played by 
Ethel Leginska), I counted the number of pixels between the centerlines
of selected perforations.  The separation for ports 0-1 was 17 pixels,
for ports 1-2 it was also 17 pixels, and for ports 0-3 it was 51
pixels.  The pixel spacing of my scanner is 0.168 mm = 0.00661 inches.

Thus the spacing between ports 0-1 was 0.112 inches, for ports 1-2 it
was also 0.112 inches, and for ports 0-3 it was 0.337 inches.  Since
0.112 inches is very close to 1/9 inch, and since 0.337 inches is close
to three times that amount, this confirms that the four leftmost ports
are spaced on 1/9 inch centers (with respect to each other) within the
accuracy of my scanner.

For the same roll I counted the number of pixels between ports 3-7, 
which I found to be 72 pixels.  (Port 6 is not used in 7510).  This
corresponds to 0.476 inches.  If the ports were on 9-per-inch centers
the separation would be ((7 - 3) / 9) = 0.444 inches, which is smaller
than the measured value by 0.032 inches.  The conclusion we draw from
this is that the four leftmost ports are offset to the left of their
expected positions by 0.032 inches.

Since 7510 is an early roll, I decided to measure Artrio-Angelus roll
8127 (Toccata from the Fifth Piano Concerto by Saint-Saens, played by
Leopold Godowsky).  This is a late roll.  There are 55 pixels between
ports 3-6 and 72 pixels between ports 3-7.  This corresponds to 0.333 +
0.031 inches for ports 3-6 and 0.444 + 0.032 inches for ports 3-7.
Thus we find that port 3 is offset by 0.032 inches to the left of its
expected position.  This matches the offset of 7510.

I cannot explain why the rolls and tracker bars have different offsets
for the four leftmost port positions.  I hope others will be able to
shed some light on this finding.

Wayne Stahnke