John Del Carlo's comments about the musicality of Dean Howe's and
Michael Swanson's recut Ampico rolls bears out what I and other roll
makers have observed -- original rolls were not always well made.
The manufacturing quality varied from very good to surprisingly poor,
even for the major brands.  Variations in tempo (step size between
punch rows) were common, particularly in the first foot or so of rolls,
as the perforators "came up to speed".

Skewing of the punches was also common.  This causes notes in the
treble, for instance, to strike earlier or later than the notes in the
bass.  Sluggish or oversensitive punches sometimes caused single notes
to be advanced or delayed from their proper positions throughout a roll.
Chaining could be erratic, though this is usually more a cosmetic
problem than an audible one.  If the chaining becomes too widely spaced
a sustained note will repeat where it shouldn't.  The laundry list
includes many things not noted here.

Years ago Wayne Stahnke developed a suite of programs that analyze the
punch positions in a roll and attempt to restore them to their intended
positions on the original timing grid.  This was successful for the
most part, though it was often necessary to make judgment calls in
ambiguous areas.  Since then others have developed similar programs.
They all do a remarkable job of putting the punches back "on grid", but
manual inspection and judgment is nearly always needed to make the new
copies as right as possible.

Wayne also developed a program, "Punch", for driving a computer
controlled roll perforator.  This provides precise digital control of
the punch positions in a roll, corresponding exactly to the restored
roll image.  The step size (tempo) can be reproduced exactly as well,
resulting in a roll that is a punch-for-punch accurate reproduction of
the original roll, but corrected to the original intent.

I make a minor adjustment to the step sizes so they correspond to the
roll being copied, rather than to the factory value.  This is for the
roll owner who compares the copy to his original so he can verify
accuracy by overlaying the two.  Paper shrinkage or expansion usually
makes any roll a different size than when manufactured.  If requested
I will use the factory value.

Fixing editing or performance errors is a controversial issue.  Where
it is a matter of artistic license it stays, whether I think it is
right or not.  On one hand, one can argue that we should preserve the
errors as an archival record, on the other hand we can argue that the
musical result is most important.

I prefer to make the roll the best listening experience possible, as
that was the original purpose.  For archival purposes the original scan
of the roll is preserved, so we always have a record of that.  Some
examples are obviously wrong notes in a chord (whether due to an artist
error or editing error), "clams" (punches that obviously don't belong),
missing punches in chains, missing Bass Crescendo punches in Ampico B
rolls, and so on.

The bottom line is that modern recuts sound at least as good as the
original rolls, and many times better.  On top of that one can play
rolls made with new paper and not worry about damaging original rolls
that are getting older and more brittle.

Dave Saul uses the same suite of programs that I do, and he has the
same philosophy of editing.  The rolls he produces are equally good.

Bob Billings - Sierra Music Rolls
Reno, Nevada, USA