In MMD 980115 I reported on Denis Hall's seminar in London on the
subject of the development of Duo-Art roll coding for the PPG.  He's
seen a transcript and contributes the following additional notes.

I had said:

> Then (c1914) H Creary Woods, the chief editor, developed the roll-
> coding console, a dark and gloomy affair with two large knobs with
> pointers on dials and two foot pedals.  Before the performance he
> would have invited the artist to mark the score up to indicate mood
> and accents.  This was propped up on the console to one side and
> while the electrical contacts on the piano made sure that note
> durations and pedalling were accurately registered on a master roll
> being cut in real time, contacts on the console registered
> accompaniment and theme dynamic codes.
>
> What the pedals did is something of a mystery but it is now believed
> that one provided additional sustaining to that provided by the
> artist and the other commanded the perforator to regard certain
> notes as "theme" notes.

Denis Hall writes:

  "Further to Dan's message, what the pedals on the recording desk
  were for is something of a mystery, but I think the two were to
  activate bass and treble "snakebites" when notes were played in the
  appropriate half of the piano.  I can't see any reason for Woods or
  Reynolds (in London) to have wanted to override the pianist's use of
  the sustaining pedal at the time of the recording; that would have
  been done during the editing if it was thought necessary.

  "It is interesting that Reynolds, working in London, always used his
  'Original' roll with the dynamic tracings made at the time of the
  recording as the basis on which to work, in the manner Woods did in
  the early days in America.  The English rolls (0---) right to the
  end of production in the early thirties always have the rapidly
  changing, sometimes apparently illogical, coding much like the
  American rolls between 5501 and, say, 5600.  Needless to say,
  Reynolds became very good at producing excellent rolls using his own
  preferred method.

  "Hofmann and Grainger are the two pianists usually quoted as being
  involved in the editing of their own rolls.  I think Harold Bauer
  was at least as involved, and started as early as 1915 to "work up"
  his interpretations.  Some of his earliest rolls are full of
  character - much more so than other pianists'.

  "Of course I picked disc/roll comparisons which would illustrate
  the similarities, but it would have been much more difficult to find
  examples where the two were _not_ very similar !

  "What goes on in the stack of a reproducing piano is still a
  mystery, and needs to be thoroughly analysed if more modern
  reproducing actions are to be designed which can give authentic
  performances from the old rolls.  A start has been made in taking
  stack dynamic level readings, and on the strength of what I have
  seen, the Accompaniment power is constantly rising and falling.
  Whether this is what the editors specifically intended or inevitable
  features of the Duo-Art system is another matter.  Whichever is the
  case, it is part and parcel of a correct reproduction of a roll."

Dan Wilson,
London