I performed this experiment a few years ago, when a library curator
inquired about archiving music rolls.  I wonder if there are other
related photographic experiments.

Abstract:

   A reproducing piano roll, 11.25 inches wide, travelling at 167 feet per
minute, was photographed by a 16-millimeter microfilm machine at 1:24
reduction.  The resulting microfilm image easily resolved the chaining
patterns but did not resolve the scalloping of the continuous slots.
Several short blurred regions were attributed to slight slipping of the
music roll.  A more precise drive mechanism is needed, and it may be
necessary to use 35-mm film.

Problem:

   Many important collections of music rolls must be archived by some
method before the music paper disintegrates from age.   Transcribing the
music roll to computer file with a pneumatic reader fails to preserve much
data of importance, for example:  the perforator advance distance (evidenced
in scalloping), the design and contents of the label, lyrics stencilled
or printed on the edge of the roll, pencilled notations, and evidence of
editing, such as adhesive tape and knife marks.

Objective:

   Identify and evaluate efficient methods of preserving music roll data
to archiving standards, with suitable precision to enable replication of
the original music roll.

Approach:

   Perform experiments with commercial microfilm machines:  photograph a
music roll and evalute the suitability of the resulting microfilm image
for subsequent optical scanning to computer disk file.

Results (July 1991):

1.  An experimental microfilm was made by Microfilming Service, Inc., of
    Corona, CA.  They are both a microfilm service bureau and retailer
    of Kodak products.

2.  The music roll was  Ampico 206511, "Sometime".  Roll condition was
    fairly good, with firm edges, length about 25 feet less trailer.  The
    label on the leader was in good condition, and the lyrics were the
    familiar Ampico green color.

3.  The piano roll was hand-fed into an old Rekordak Rotoline Microfilmer,
    Model RD-3.  This machine has a fixed 24:1 reduction ratio (standard)
    and transports the original at 167 linear feet of paper per minute.

4.  The film used was Kodak Imagelink FS Microfilm 1455, Estar Base,
    16mm x 100 ft.,  Film Spec. No. 615.  This emulsion is similar to the
    obsolete 1454 fast-speed emulation,  with better contrast.  The plastic
    spool is to an ANSI standard.

5.  After the film was developed a few prints were made using an old office
    microfilm pviewer/printer onto direct-exposure silver-halide paper.
    Comparison prints also were made at the same image size (85%) with an
    office copier (direct optical xerographic, not a scanner & laser printer).

6.  The image of the roll width on the microfilm was measured at
    0.469 +/- .003 inches, which compares favorably with the computed
    size of 11.25/24 = 0.468 inches.  The reduction along the "long" axis
    wasn't measured.

7.  The microfilm failed to resolve the little points of scalloping.  Also,
    the roll took a jump a few times of about 0.180 inch --  one blurred
    region is visible just below the pencilled "Paderewski" notation, and
    there are others of varying degree throughout the roll.

8.  The reduction ratio of 1:24 is the standard for all Kodak 16-mm
    microfilm cameras;  it's use is recommended.

9.  The green lyrics appear fuzzier than the hole images, possibly due
    to chromatic aberration.

10. The sales staff at the service company recommended these other Kodak
    products for further experiments (unfortunately, none were readily
    available at the time):

    Kodak Mini Microfilmer --  portable16mm, 12-inch field width, 1:25 nominal
    ratio, resolution 100 line-pairs per millimeter minimum, object speed
    10 to 15 feet per minute.  Price $5000.

    Kodak Recordak Rotoline Microfilmer Model RD-3 -- 1:24 or 1:32 reduction
    options, 167 feet per minute.  (This is the machine used for the above
    experiment.)

    Kodak Imagelink FS Microfilm 1455, Estar Base, 16mm x 100 ft.,  Film
    Spec. No. 615 -- price about $9 per 100-foot roll, including processing.

11. Another division of Kodak handles optical scanning machines which
    create CCITT Group 3 or 4 formatted images at speeds of 18 inches
    per second across a 12-inch field.

Recommendations:

   The Model RD-3 machine used in this experiment was admittedly old, and
it's probable that a machine in better mechanical condition would produce
a sharper film.  It would be worthwhile locating a machine in better
condition and repeating the experiment.  It shows good promise.

-- Robbie Rhodes    13 August 1995