I imagine Eugene Rogers will get several responses to his query asking
how music rolls were produced, each approaching the question from a
slightly different angle.  Here's mine.

The Leabarjan perforator CAN be used to perforate a single roll, and that
roll CAN be used as a master from which multiple copies may be made using
a production perforator -- a perforator capable of punching multiple copies
(by punching through multiple layers of paper, a dozen or maybe two dozen
layers at a time) and capable of punching the paper at a fast clip,
perhaps a foot a minute, as it reads the master.  This the Leabarjan is
in no way capable of doing!  At least not the little Leabarjan which the
company sold with high hopes that it would find use in school music
classes.  I wonder whether it ever did?

L. Douglas Henderson produces the master rolls for his Artcraft pianola
rolls using a Leabarjan like Eugene was reading about.  He then auditions
and edits the roll by playing the hand-made master on his studio pianos.
When the master has been edited to his satisfaction, he sends it to a
company which operates production perforators in order to get the
requisite number of copies made.  Nobody, not even a zealot like Doug,
would think of punching out sale copies on a Leabarjan.

I don't know whether Leabarjan ever made production perforators or not.
The most famous brand of production perforator is the Acme.  QRS Music
Company in Buffalo uses double-gang perforators, which may have been
designed especially for QRS way back in the early days by Ernest Clark
(or was it Melville?).  The tracker and the pneumatics for reading the
master sit in the middle of the setup, while to the left and the right
are twin perforators driven by the one master reader, giving QRS an
unusually high production capacity per machine operator.

Because of the mechanics of copying from a master, too many errors are
introduced into the copies by the copying process if the master is made
to the same scale as the copies.  So the perforations in the master are
made longer (and larger too) than those in the final copies.  Wurlitzer
band organ masters were made on a 3-to-1 scale: that is, the master
perforations were three times as long as the finished perforations needed
to be.

There are basically two methods of getting the music from staff notation
form to holes in the master.  The one most often used, the "drafting
board" method, simply involves someone marking out the tune by hand on
the master paper, note by note, measure by measure.  The method is slow
but sure, and if the arranger is gifted the results don't sound too
mechanical.

The other method, the "hand played" method, involves a musician sitting
down at a musical instrument -- piano or organ or whatever -- but one
specially equipped with sensors and recorders which will record on paper
or other medium the notes struck (and for reproducing pianos, the
intensity (velocity) with which each is struck, as well).  From that
record a master is produced.   But experience shows that the "hand
played" master still needs some editing -- often a lot of it -- to smooth
out the performance.  A debate still rages today over the authenticity of
some "hand played" music rolls, as well as over the virtues and defects
of hand played versus drafting board arrangements.

Reproducing piano rolls are apt to come from hand-played masters.  Band
organ rolls never did, to my knowledge.  (I have a photograph of Ralph
Tussing sitting at his arranging table -- -one once used in the Wurlitzer
factory -- -with sheet music propped up in front of him, leaning over the
master paper, using pencil, triangle and T-square to mark out the notes
for a band organ master.)  Player piano rolls are probably somewhere in
the middle, though player piano roll labels have to be taken with a grain
of salt with regard to "hand played" claims as well as with regard to the
performers credited with the "playing."

After the master was marked, no matter how the marking was accomplished,
it was left to the ladies in the factory to punch out the notes where
marked on the master.  Then after auditioning and clean-up editing, the
master was ready to drive a production perforator in making copies for
sale.

The QRS factory (too bad Eugene is down in Florida; he'd enjoy a factory
tour so see what I have tried to describe here) has both a marking piano
and the ingenious direct-master-punching piano which J. Lawrence Cook
"played" in making many of the QRS rolls bearing his name or the name of
the performer whose style he imitated.  I say "played," because it was a
slow-motion stop-and-go, chord-by-chord process you had to master in
order to operate that master-punching piano.  It wasn't played; it was
driven.

Today QRS's arranger, Rudy Martin, uses computer software to do his
arranging on.  I am not sure how Rudy gets from computer disk to master,
however.  Beyond that, a piano equipped with Wayne Stahnke's marvelous
solenoids and MIDI capability, such as the Bosendorfer at IPAM
(International Piano Archives at Maryland), takes the science or art of
reproducing a piano performance into a whole new realm, never dreamt of
by Welte, Stoddard or Hickman.