Lester Hawksby wrote: "... However, surely if when the roll is punched
it is taken up onto a spool of the same diameter as the player piano's,
the note length and hole spacing when it's being punched will increase
by the right amount to make the roll play correctly at constant Tempo?
(Regardless of speed, the increase of linear distance with increase of
diameter is the same)."

My experience is only with two perforators built by others, but I would
_not_ recommend the approach to pull the paper through the punching
machine via the take-up spool.  It might (and will, in my experience)
happen that the paper builds up, but at some point, the resistance of
the paper in the machine might become somewhat larger (for whatever
reason; see below), and therefore, the rotation of the take-up spool
will just coil together the rolled-up paper more tightly, which will
therefore _not advance at all_.

Only when the paper is sufficiently compressed will its outer end start
to move again.  Because of the general problem of friction that sticking
friction is larger than sliding friction, the change from stopped paper
to moving paper is totally uncontrollable, and hence you will not get
continuously moving paper in this case.

In a way, pulling the paper through the machine with the take-up spool
is a mechanically quite indirect, and hence error-prone, idea.  The
take-up spool pulls at the _head_ of the paper, which then transfers
the force through the _friction of many layers_ on that roll to the
current _tail_, where you need the pulling force.  Thus, there are many
uncontrollable variables that influence the quality of that complex
process - typically a sign that the process will fail in random ways.

The main reason for the problem is with a simple-mindedly designed
machine, where the take-up spool must _also_ pull the paper off the
(usually large and heavy) paper supply roll.  Howeer, this could by
remedied with a separate motor to turn that roll so that paper is
always fed loosely to the punching machine.

Still, the punching machine internally needs (I think) a certain amount
of friction, just to keep the paper flat -- thus it must be pulled back,
maybe _just a tiny bit_, before the paper is punched.  But even this
tiny friction might at some point, with much wound-up paper on the
take-up spool, momentarily be larger than the current force exerted by
the wound-up paper layers, and the paper ... will stop!

You'll pull your hair out if that happens 20 or 40 or 60 feet into
a roll for the first time, because you have to throw away the whole
roll then.  And you will sit with fear and horror in front of your
machine at every inch of every roll you punch in the future whether
it will happen _just now_.  Especially the last foot will give you
a heart-race about whether the roll will survive it -- the paper is
the longest now on the take-up spool with the most layers, so it is
most uncontrollable in its frictional behavior, but on the other hand
you have the most to lose just now, namely the whole roll.

Lastly, the idea that the build-up on the take-up spool of the punching
machine mirrors that on the player piano prevents punching multiple
rolls in one run: you have to get up at 3 in the night to get out the
punched roll and hook up the paper anew for the next roll.  The faster
your punching machine is, the less sleep you will have ... or so.  For
this reason alone I cannot imagine that the professionals of the older
days did it like that (unless someone invented the super-duper
punched-roll changer).

For my own sanity (both with the nightly sleep, but more so because of
the fear and horror), I'd find a better solution ... which is _directly
pulling_the paper immediately after the punch with rolls with some sort
of rubber tires (a handful of tires spread out over the width of the
paper is enough). This avoids the very indirect way of pulling the
paper via the take-up spool described above.

For the tempo compensation problem, you then have essentially three
possible solutions:

a) Don't compensate the build-up -- i.e., assume that the take-up spools
of the instruments where the rolls are played are "sufficiently large",
and/or the paper is sufficiently thin.

b) Compensate with punching length -- i.e., instruct your software to
punch longer hole rows towards the end.

c) In theory, you could also compensate by increasing the paper speed
at the pulling rolls -- i.e., install some software there that drives
the motors (e.g., stepper motors) faster and faster.  But this will
increase the hole distances, which might or might not be a good idea.
I do not think that anyone does or did it like that.

So much for my 3 cents -- mainly armchair, but also gained from the
experience of punching rolls for 20 years for a multitude of small
organs and, lately and experimentally, for a player piano (where we
do not use any tempo compensation).

Harald Mueller