This note is intended for those contributors to the Digest who requested
a copy of the pattern roll package I offered a few days ago.  By now, all
of the packages have been sent and received.

In addition to the "View" program, the package contains the restored
pattern roll for Ampico roll 57504, Rachmaninoff's Prelude in C-sharp
minor, Op. 3, No. 2, played by the composer.

In order to acquaint yourself with the concept of a pattern
roll, I suggest one or more of the following exercises:

    (1) Ampico rolls of this period have 98 ports across the tracker
bar, numbered 1 through 98 from bass to treble.  We know from the test
roll (Ampico roll 61391) that there are exactly 356 rows per foot along
the length of the roll.  Mark off a checkerboard of 98 by 356 squares on
graph paper, each square representing one punch position.  Reconstruct
the pattern roll by hand for the first foot (304.8 mm) of the production
roll, using a magnifying glass and straightedge.  This is easier than it
sounds, and should take less than an hour even for a person who has never
attempted it before.  Be sure to reconstruct the number of "empty" rows
in the white space exactly.  Watch out for bass and treble expression
that appears to be in the same row, but is actually offset by one row.
When you are done, compare your reconstructed image with the one provided
in the package, and resolve any differences.

    (2) The webbing, or bridging, was accomplished by using a special
pattern, depending on the channel.  In the note field, this pattern is
punch-skip-skip-punch.  Find the webbing patterns for all of the channels
in the roll.

    (3) Immediately following the three octaves at the start (A, G#, C#)
are three chords that span the stack division that falls between E and F
above middle C.  The top note of these chords is intended to speak more
loudly than the remaining notes.  This is accomplished by a special
coding trick, characteristic of Ampico rolls of this period, that relies
on locating a cancel in a particular row.  Examine five or ten examples
of this coding trick, and (if you have the patience) write an article for
the Digest about what it is and how it works.

    (4) A few years later, the Ampico coders gave up using this trick and
started using a different one.  During this later period, Ampico roll
57504 was re-coded and the new trick was substituted.  The exercise is to
locate a copy of this later roll, examine it, and provide an article to
the Digest that compares and contrasts the two different approaches,
along with a discussion of the reasons for the change.

There are many more possible exercises that naturally present themselves,
and these will be obvious to the user.  I hope that one or more readers
of the Digest will go to the trouble of providing detailed articles for
the Digest explaining what they have learned about the practices of the
Ampico roll coders.

Wayne Stahnke


 [ Editor's note:
 [
 [ Wayne has identified 3 different versions of Ampico 57504 which were
 [ issued by Ampico.  Unfortunately, two versions bear the same number!
 [
 [ "Version A" is the first issue, released in 1921, and it is found
 [ in most collections today.  Edgar Fairchild said that he did all the
 [ editing and expression coding.

 [ "Version B" is quite rare; Wayne has located only two copies thus far,
 [ and he hopes the collectors can find more.  This version was issued
 [ about 1927, after Ampico abandoned the editing practice of adding
 [ note extensions, and so the music roll appears much more like a
 [ "live recording".  Besides the obvious absence of the note-extensions,
 [ this version can be identified by the Trigger Cancel hole in channel
 [ 0B at the start of the roll.
 [
 [ "Version X" of the song is included in jumbo roll 100075, issued in
 [ Nov. 1929, and it's different from the previous two versions!
 [
 [ -- Robbie