When we copy rolls, we must address the problem of accuracy. This issue
is difficult to think about, and somewhat arcane; perhaps for these
reasons, many people have historically not thought about it at all, and
instead have simply assumed that any copy is a perfect one, like copying
a file from one disk to another in a computer. It is easy to demonstrate
that this is not the case; copies vary in quality from excellent to truly
awful. The copies currently made by Custom Music Rolls, as an example,
are of very high quality and perform extremely well; however, some older
copies made on other equipment were so bad that their owners have told me
that they finally discarded them in disgust.
The fundamental problem is as follows. Music rolls are perforated in
discrete "samples," or rows. Given a production music roll, examination
shows that the holes in the roll do not fall at any position along the
length of the roll, but only in discrete positions, all of them multiples
of the perforator advance. Thus, each perforation falls either in one
rows or the next; holes cannot fall in the cracks. This is somewhat
analogous to playing the piano; one can depress one key or the adjacent
one, but there is no key in between.
When we go about copying rolls, we can either consider this structure or
ignore it. If we ignore it, our copies are approximations, however, good
they may be. This approach can yield excellent results in the hands of a
skilled practitioner. Julian Dyer and Bob Billings address a different
question, however, which I understand to be this: given our druthers,
how would we copy rolls?
To my mind, the model for what we would like to do is furnished by the
example provided by modern Ampico rolls. By dint of extremely good
fortune (and a good deal of effort by Larry Givens) the original Ampico
perforators and many of the Ampico master rolls survive. Over the past
20 years or so, there have been new copies made on this original
equipment, using modern paper. These copies are excellent, and do not
exhibit the problems we are accustomed to with historical rolls. The
paper is flat and straight; the individual holes are round and flat, and
the rolls track and play wonderfully. In making rolls on this equipment,
we are returning to the intent of the originators of the rolls in the
1920s. The new rolls are hole-for-hole copies, made anew.
I think that we should set as our long-term goal achieving the sort of
accuracy and integrity that we see with modern Ampico rolls. Doing this
involves two intermediate steps. First, instead of simply copying rolls
as we find them, we must use the rolls that have come down to us as the
templates for reconstructing, as best we can, the master rolls from which
they are derived. Second, we must have perforating equipment that makes
use of the restored master rolls to create "new original" rolls. Only in
this way can be realize the intent of the originators of the rolls, and
produce copies worthy of their efforts.
Fortunately for use, the internal structure of existing production rolls
allows us to recreate the master roll perfectly in many cases. Discrete
rows in production rolls provide us with what we need to reconstruct the
sprocket holes in the master roll; in other words, it is possible to tell
which row of the master roll each perforation was derived from. If the
original rolls were not sampled, this would not be the case.
Thus, the original sampling of the data, although done originally for
reasons entirely unrelated to our attempts to copy rolls, provides us
with a Rosetta stone. The sampled structure allows us to recreate the
master rolls, and (following that) exactly replica rolls. This is an
extremely fortunate circumstance. I do not know of any parallel with
other historical documents, such as sound recordings.
To my mind, we should set as our goal creating new original rolls made
from restored master rolls. This goal, although difficult to achieve and
currently uneconomical, is the best way of preserving the legacy that has
been handed down to us.
With best regards, I remain,