George Bogatko raises some good points about how one stores and reads
data.  Storing the music data on paper, as he suggests, is certainly
a good idea, presuming one has the space to store thousands of rolls.
Since most of us don't, we have to make compromises.  Storing the data
on digital media does present the problem of longevity and future data
retrieval, but there are ways around this.

Since the days when I naively recorded rare records on a wire recorder
I have been sensitive to degradation of media.  My solution is to
transfer data to the latest format, keeping the old "just in case".
Copying digital data regularly to even the same media provides some
protection against degradation of the original media.  I have data
originally stored on 5-1/2 floppies that now resides on CDs and is
still quite happy, thank you.  And he who doesn't backup regularly
has only himself to blame when (not if) he gets skewered.

Wayne Stahnke was the first to use the "phase lock" scheme to recreate
the original roll master pattern, which is the Holy Grail of reading
rolls.  Others are aware of the need to phase lock, and are working on
methods to do so now, present company included.  Even rolls read with
quantizing errors should yield phase locked versions, given the proper
software, and as long as the quantizing error is less than one-half
the original roll step.  Current CIS readers have quite good enough
resolution to provide files that can be easily converted to phase
locked ones.  Line scan cameras can also record gray scale, which
allows capture of printed matter on the rolls.  Converting that data
to something useful is another matter....

Paper advance is certainly an issue for readers.  Capstan advance is
the obvious choice for creating an exact image of the roll.  A method
I've experimented with is using a digital encoder riding on the roll to
add position data to the line data.  Theoretically, if the phase lock
loop has enough range, it could lock to a roll advanced by the take-up
spool.  Capstan advance by a stepper motor is the best choice, in my
estimation.

Wayne's software suite for creating rolls is superb.  Besides all
the pre-punching utilities he provides, his Punch program controls
perforating as close as one wants or needs (and far beyond that,
I might add).  Adding stepper motor control to Punch has allowed
recreating the original perforator advance to the accuracy of the
stepper advance.  Careful choice of the stepper advance minimizes the
error for any of the known original advances.  With my perforator,
the maximum error is less than 0.001 inch.  Plus the inevitable phase
noise, of course.

With phase locked masters, punch size does not matter as far as
the stored data goes.  It only becomes significant when the roll
is actually punched and the holes must match the tracker bar ports.
Other than that, the time relationship of the roll holes to each other
is the only thing that really matters.

More power to all the guys working on small and portable scanners.
There are many collectors who might allow someone to visit and read
rolls, who otherwise refuse to let rolls out of their hands.

Bob Billings