I am gratified by the response to my note in this place asking for
help in determining if certain Artrio-Angelus rolls have Mel-O-Dee
twins [180810 MMDigest].  The feedback has been very helpful, and with
the new information I received I have able to make some progress.
This note is a summary of what I learned.

Terry Smythe generously sent me a link to high-quality scans he made
of four Mel-O-Dee catalogs.  These will be a valuable resource going
forward.

At Anthony Robinson's suggestion, I examined the list of Mel-O-Dee
rolls on Robert Perry's web site, https://pianola.co.nz/public/ 
On comparing that list with the Artrio-Angelus rolls available to me
I found that Robert has scanned four rolls that are possible twins,
all played by Edythe Baker.  These are

     4019 "Avalon" (Jolson/Baker),
     5125 "Somebody's Wrong" (Marshall-Egan-Whiting/Baker),
     5126 "Steamboat Sal" (Fisher-Raskin-Causer/Baker),
     5465 "12th Street Blues" (Heagney/Baker),

for which the corresponding Artrio-Angelus roll numbers are 2019,
2125-D, 2123-D, and 2273 respectively.  (Many Artrio-Angelus rolls
bearing the "-D" suffix have Duo-Art twins, but that is a subject for
a different note.)  All four of these rolls have the corresponding
Mel-O-Dee roll numbers stenciled onto the paper just before the rewind
perforations.

For two of the four titles, the roll numbers differ only in the first
digit (4019/2019, 5125/2125-D).  It is not clear if this is a
coincidence or if some of the roll numbers were chosen for parallelism.

I examined the Mel-O-Dee catalogs to see what I could learn about
Robert's rolls.  To my surprise, only "12th Street Blues" appears in
any of them.  The September, 1920 catalog and the 1923 catalog both
include it and give its roll number as 3893, different from the number
of Robert's roll.

I emailed Robert, who kindly provided me with the scans.  They are
of sufficiently high quality that I was able to reconstruct punch
matrices for all of them.  This allows a direct comparison with the
reconstructed punch matrices for the Artrio-Angelus rolls.  On making
the comparisons, I found that the rolls are indeed twins.

The Mel-O-Dee rolls lack expression coding, of course, but otherwise
the pairs differ in only very small ways:

1. One difference is single-row errors, by which I mean that the onset
or release of a perforation is one row early or late.  It is easy to
imagine that differences of this type are caused by perforator errors.

2. Another difference is in the webbing.  Long slots in Mel-O-Dee rolls
are sometimes broken up by inserting webbing in the Artrio-Angelus
issues.

3. Finally, in quite a few cases perforations in the Artrio-Angelus
rolls have been extended, resulting in a more legato effect.  This
indicates that the same master roll was not in fact used for twins.

Which roll was derived from which?  Because some Artrio-Angelus rolls
(2273 is among them) are labeled "Melodee Music Co. License" on both
the roll label and box label, I assume that the original master roll
came from Mel-O-Dee and was subsequently reworked for use with
Artrio-Angelus.

Since the pitch of all of these rolls is 1.20 mm = 1/254 foot, twins
are of the same length.  Thus a single word stencil would serve for
both.

We can not conclude that the same stencil was used for both issues,
but this discovery shows it was very likely.  In other words, what
we see in these four twins is consistent with the use of a single
stencil for each pair.

Wayne Stahnke
Reno, Nevada