Spencer Chase is curious as to the resolution a perforator should have
to copy all kinds of old rolls and to turn out an accurate reproduction.
I would like to enter this thread and add my two cents for what it's worth.

I'm not a roll cutting guru by any means, and there may be some others
here who would like to correct this opinion, which is why I am writing it.
Here is my opinion about roll accuracy, as it relates to the most critical
of all masters, that of the Ampico reproducing piano, whose masters were
actually cut 3 to 1 in length to be as precise as possible (most other
rolls were cut 2 to 1).

The most single important measurement to maintain in a high quality
reproducer roll is the perforation width. A wide hole triggers each note
or intensity with minimum time variations and greatly narrows the natural
differences between valves and sensitivities created by a host of
uncontrollable factors. The best rolls had holes 5/64" or perhaps a bit
wider, from my measurements. (I can slide a .081 drill bit through the
holes of an original Ampico roll, for example, even though the trackerbar
hole measures only .072 wide. That leaves about .039" for each web between
the notes). The reason I stress punch diameter is because perforators with
smaller punches do not perform well when the most sensitive notes are to
be played. For regular player pianos, this is not a problem. But when
trying to play consistently at 5" of vacuum or below, it becomes a very
big problem. Unless the holes are correctly sized, the roll won't play
as well regardless of stepping increments.

Now, for resolution, let's say your punches are .081" diam. (#46) and you
wish to make a trial run with a perforation every .081" on the note sheet.
That gives you a perf chain without any distance between them. You would
get 148.148 holes/foot. Granted, your resolution has to be better than
that, because your punch is .081 and you will need to be able to make at
least three steps of the perforator in that distance. Four steps seems to
me to be adequate for anything you would ever do in paper. That means that
each step equals about .020.

Here is one way to arrive at this number: The height of a trackerbar hole
is .060. The time it would take a .020 increment to pass the hole at a
tempo of 70 is: .020+.081~ .100 , .100 divided by 1.4 in/sec = .07
seconds. That is the minimum length of time, or time increment that you
are able to differentiate between in your music at this tempo. Of course,
this allows for step differences between different makes of rolls too
because the player itself could not possibly resolve a chain of increments
like that with only .020 bridge between them, but you are also interested
in precisely copying a roll that didn't use a multiple of your steps. (The
difference between four-steps per diameter and three steps per diameter is
.02 seconds, or .027 inches per step difference.)

Frankly, for a number of reasons, I prefer to divide the diameter of the
punch by an odd number of steps. It would seem to me that if you were to
divide the .081 diameter by 5, or .0162"/step, giving 61.7 steps/ inch,
this would seem to me to be workable, leaving .4 step over at the end of
each foot. That seems to me you can never miss a punch by much over .2
step increment (about .0003), worst case. So about three times your
postulated basic increment of .005" would be about perfect, and easier to
do. If you want to use even numbers, you could double your .005 to .010
and do it just as well, I would think.

The wisdom of step increments which are not even numbers per foot seems to
me to allow a perforator to be more accurate by seldom falling into a
"slot" in which the start or end of a hole varies across its own step
lines more than 25%, and so is resolved to the closest step line.
Sometimes, over-punching or missing the first punch by a tiny fraction
is caused whenever music is laid out by the foot and then changes a little
over time.

The problems of perforator error, as I have understood it, came in trying
to do it with even number increments/foot and /or even number of step
lengths/ft. Since the perforator mechanisms copied technology developed
first in the printing trade, one might go back to the incrementing used
historically (if such a thing existed for one reason or another) to see
if he could find some common scale. It's a sure thing that the 88 note
trackerbar for player pianos was developed from the pica scale used by
printers, and scale engines could be set up to graduate just about
anything, English, Metric, or Timbuktu.

 [ I, too, discovered the 9/inch printers scale when I worked at a print
 [ shop.  (I think it was named "Agate".)  But Craig, is it really a "sure
 [ thing" that the music roll industry specifically adopted the printers
 [ standard?  After all, in that era the decimal rule was very uncommon,
 [ and many industries defined their own independent standards.  Can you
 [ offer evidence for your statement?  -- Robbie

Craig Brougher