DuoArt Roll Punch Advance
By Wayne Stahnke
This note is intended primarily for Spencer Chase, but is presented through the Digest for the benefit of other readers with an interest in the subject.
Spencer, you have raised an interesting question that has never been satisfactorily investigated or answered. The punch advance (also called step increment) of rolls made during the 1920s is a mystery to all of us who have studied it.
The natural tendency is to assume that the punch advance is a "simple" number. We know that in every period designers seem to feel an almost overwhelming pressure to build new instruments in such a way that the target dimensions are "round" numbers, in the sense of being whole increments of the smallest unit of measure commonly in use. This is not just a consequence of an exaggerated sense of orderliness; it is also convenient at any time and place to use simple numbers that are easily manipulated and measured in the workshop. This tendency can sometimes be used to reconstruct lost standards. When measurements of any device show its dimensions to be very close to a "round" number, it is generally safe to assume that the whole number is the intended dimension.
This sort of argument can be used as follows. We find that for 88note rolls in general, and for QRS rolls of the early 1920s in particular, the punch advance is very close to 240 punch steps per foot. Such rolls usually (but not always) have exactly 24 punch advances per beat; this lends itself to subdivision into sixteenth notes, thirtysecond notes, and triplets, all without roundoff error.
Consider the timing that pertains at Tempo 100, corresponding (by definition) to a paper velocity of 10 feet per minute. At 240 punch steps per foot there are exactly 2400 punch steps per minute. One beat occupies 24 punch steps, so at Tempo 100 there are exactly 100 beats per minutethe marking on Maelzel's Metronome for Tempo 100! Similarly, Tempo 80 corresponds to M.M. = 80, and so forth. It seems (to me at least) that this argues strongly in favor of concluding that the perforator designers chose 240 punch steps per foot as the target punch advance, so that the "Tempo" marking on the roll matches the musical tempo, as shown on the standard metronome.
Given this approach, we would expect that the punch advance of reproducing piano rolls would similarly be an "even" number of punch steps per foot. It is therefore surprising to learn that it is not. As you pointed out, the measured punch advance is usually close to a "round" number, but not close enough to allow the discrepancy to be attributed either to manufacturing tolerances or experimental error, or to any combination of the two.
Moreover, there is solid evidence that "round" numbers are not the intended values at all. As an example of this, consider Ampico Roll No. 61391. This roll contains a Tempo Test in which chords have apparently been manually inserted every 6 inches. A reconstruction of the master rolls shows that these chords do not appear every 180 punch steps, as one would expect, given that the punch advance is nearly 360 punch steps per foot. If memory serves, the timing chords are spaced exactly 178 punch steps apart, uniformly from one to the next throughout the timing test, yielding a punch advance of exactly 356 punch steps per foot. We are extremely lucky that in this case we can determine the punch advance exactly.
Similarly, late WelteMignon Licensee rolls exhibit a punch advance of about 0.756 mm. (Early rolls use a coarser advance.) This is close to 400 punch steps per foot, but it is enough different (almost 1 percent), and so marvelously repeatable from roll to roll, that the intended value is certainly different. I have yet to determine what that intended value is.
In your note, you mentioned that your immediate interest is in DuoArt rolls. The punch advance of these rolls was changed sometime in the mid 1920s. Early rolls have an advance close to 256 punch steps per foot (or 3/64 inch per step). Late rolls have an advance almost exactly 11/2 times this value, or 384 punch steps per foot. I do not know the exact numbers. However, I have been able to locate, scan, and reconstruct the master roll for the 1921 DuoArt Test Roll. (I undertook this work at the suggestion of John Grant, who generously provided an original roll from the period for the purpose.) My recollection is that this test roll, like Ampico Roll 61391, provides the exact value, competely without equivocation. I am writing this note from home, so I do not have the value at hand.
I hope this note has been useful. It is my intent to encourage you and others to study this problem. I have great hopes that some day we will have a complete knowledge of this and similar issues related to the perforator technology of the 1920s.
With best regards,
Wayne Stahnke

(Message sent Tue 3 Dec 1996, 05:36:17 GMT, from time zone GMT0500.) 

