One of the reasons John Farrell's rolls sound so good is that the
rolls are made with twice the resolution of the original nickelodeon
rolls, and he used that in his arranging.  He would often move note
timing slightly off beat to sound more like the playing of a live
pianist or to make a note stand out from the rest of a chord (known
in the trade as "agogics").  With the finer resolution the effects
were more subtle and lifelike than could be gotten with the original
resolution.

With all due respect to the old nickelodeon roll makers, they were
selling into a market that wasn't critical about the sound of the
rolls, as long as they weren't too bad.  I've copied a lot of original
rolls and found that many were not made well, and sometimes even the
editing was sloppy.  This is easy to see with my copying process.

After scanning a roll the resulting file is run through a program that
puts the punches on the original grid.  A trial copy is then overlaid
on the original for checking.  This is where the flaws in the originals
show up; punches not on beat, skewing of the paper, step size changes,
missed punches, to just name a few.  After all the physical problems
have been resolved I listen to the file via a computer and verify that
it sounds right.  Obvious problems are fixed at this time.  Even if
there are no problems found the copy often sounds better than the
original because the punches are cleaner and the notes are all on grid.

One problem often shows up in otherwise perfect original rolls:
slow drift of step size in a tune.  Usually this is at the beginning
of a tune, where the tempo is slower and gradually comes up to speed
in a foot or so.  Original nickelodeon rolls were made one tune at
a time from a library of single tunes.  The perforator was stopped
between each tune for the master change, then restarted, sometimes at
a different step size.  Apparently the perforator/master coupling had
slop or slippage in these cases.

Occasionally there would be step size drift in the middle of a tune
as well.  The roll would have a fixed step size and then slowly change
larger or smaller, then go back to the original step size.  The
interesting thing is that when the step size went back to normal the
following notes were on the same grid as the notes before the drift.
This seems to indicate some sort of elastic slippage that eventually
got back to normal.

There's a lot more that can be said when looking at rolls from the
trenches, but that might be material for an AMICA workshop.

Bob Billings