-- forwarded message, please reply to sender and MMD ---

Response to Douglas Henderson

I thank Doug Henderson for his lengthy review of the recent performance
of Antheil's Ballet Mecanique in Lowell, Mass., and while I don't want
to get into a debate with him over some of the fine points of the
history and technical challenges of the Ballet Mecanique -- especially
inasmuch as he seems to have access to information that I have not been
able to find elsewhere -- I would like to raise a few factual issues,
both to set the record straight, and also in hopes perhaps of keeping
some "myths" that have grown up around the piece from spreading
further.  And since there are some aspects of this piece that are still
unclear to me, after a year and a half of intensive research, perhaps
by raising some of these questions together, we might be able to find
some answers.

> The first part of the program showcased Jeffrey Fischer, conductor of
> the University of Massachusetts Lowell Percussion ensemble, with piano
> soloists Juanita Tsu and John McDonald.

Actually, Juanita and John were in the audience the entire first half,
their services not being required until the Ballet Mecanique.

> I never learned any more about the differences between the Grayson
> performances and the others, since Mark got interrupted and others
> came to view my player rolls, one of which belonged to George
> Antheil, as stated above.

I was a little disconcerted to hear that someone purchased some rolls
from Douglas out in the lobby, inasmuch as this was strictly a non-
commercial event: tickets were free and we weren't even allowed to sell
popcorn.

> The finale from the Mendelssohn 'Italian' symphony represented a lot
> of work, obviously, since it involved so many electronic player
> instruments.  However Mark, and other musicians I knew in the
> audience that night, told me that this was locked too much into
> notation to suit their fancy.  I felt that my Mendelssohn should be
> frothy, with defined accents, and that the trills shouldn't come on
> like machine guns when they were required.  Thus, any "Gatling gun
> trill" is a performance 'no-no' in my book.

It is in mine too, which is why every note of the piece was played in
by hand, more or less in real time.  As a long-time orchestral player
myself, and as a leading proponent of and authority on expression in
MIDI-based music, I would never wish to foist on any audience a
"notation-locked" performance -- by which I assume Douglas means one
which is "step-timed" into the sequencer -- unless of course it is
specifically called for, as in the case of the Ballet Mecanique.

It's interesting that Douglas didn't recognize this major difference
between the two pieces, and perhaps instructive of how his (and maybe
other audience members') ears respond to trills on a piano -- that
they're really not the same as trills on an orchestral instrument.
It's also quite possible that trills at extremely high velocity on
a piano sound unnatural because, in fact, they are -- human pianists
cannot play trills as loudly as they can other notes.

By the way, the instrumentalists, composers, and conductors in the
audience that I spoke to had the opposite reaction -- they expressed
amazement that the pianos could be so expressive.  But I guess it all
depends on your expectations.

> The answer lies in the fact that the more pianos one adds, the more
> the strings cancel the sound, acoustics being what they are.  Igor
> Stravinsky didn't understand this (and I have a book which details
> his letters with the Pleyel roll factory during the arranging of
> rolls in his name).  What Igor did, Antheil copied.

I'm afraid this is rubbish.  Since none of the parts of the Mendelssohn
were copied from one track to another, but instead were individually
recorded, there were of necessity small but significant variations
among what the different pianos were playing, and to say that they
"canceled each other out" is akin to saying that orchestral string
sections shouldn't contain more than four players since more than that
will "cancel each other out".  And furthermore, at the last minute,
I did decide to double all the piano parts, and play them on 16 instead
of the original eight.  The increase in volume was significant, and
very satisfying.

> Mark Lutton ... had a key seat in Durgin Hall.  Being a MIDI expert,
> and also knowing its many limitations, he thought there might be a
> 500-millisecond 'delay' with that many pianos running on a program.
> This he called a "MIDI smear", and upon talking with him during
> intermission, he detailed the time span between the instruments on
> each side of the auditorium.  Clearly, the Disklaviers, even from
> where we sat, were not truly synchronized, as claimed in the concert
> publicity.

Douglas's interpretation of Mark's statement is unfortunately way off
the 'Mark', if you'll pardon the pun.  The Yamaha Disklaviers do indeed
have a 500 msec delay when they receive outside MIDI data, but (and
I tested this timing issue thoroughly in my preparation of the files)
this delay is absolutely precise and consistent from note to note and
from instrument to instrument (and there's a very strong design deci-
sion behind it), and thus the amount of "smear" is actually _less_
than in a typical MIDI synthesizer, where the CPU is expected to
respond to incoming data instantaneously.

Furthermore, we used throughout the concert a multiport MIDI interface
running in "fast" mode, which allowed us to use in the Mendelssohn four
distinct MIDI cables with only two piano parts on each, and in the
Antheil, *one* MIDI piano part per cable, and thus avoid any potential
MIDI bottlenecks.  I suspect that whatever smear was being heard was
due to the physical distance between the pianos: remember, sound takes
about 1 millisecond to travel one foot, so if the pianos are spread out
over a 30-foot proscenium, you're going to hear interesting effects of
all kinds.

> I, for one, would have liked to know more about this mysterious
> "Pleyel patent" that was supposed to synchronize Pianolas, which
> Mr. Lehrman mentioned in his entertaining dialogue, interspersed
> with slides and audio clips related to the composer.

As would I.  I was told about it by Rex Lawson, and it was confirmed
by Denis Hall, but I have been unable to actually see it.  One assumes
it's buried in the French patent office, and my friends in France tell
me it would take a month of daily visits to get it released for exam-
ination.  So although I have no primary evidence of its existence, it
fits in extremely well with everything else I know about the early days
of the piece: that Antheil intended to perform it on 16 synchronized
Pleyelas, and that he had to come up with all kinds of excuses why he
couldn't do it (Pleyel was too slow cutting the rolls, there wasn't
enough electricity in any hall in Paris to run 16 Pleyelas, he was
eaten by lions in Africa, that sort of thing) until he finally did it
with one Pleyela and a large number of overworked human pianists.

And I'm interested as to why Douglas referred to my remarks as a
"dialogue" -- was there someone else on the stage I wasn't aware of?

> While the audience was told that 16 pneumatic mechanical pianos
> couldn't be synchronized until now, with the emergence of the
> Disklavier, this is not true...

I am grateful to Douglas for his extensive list of synchronizable
player pianos, most of which I was not previously aware of, but let's
face it: these were all historical rarities, and I think it's safe to
assume that it was never, nor would it ever be, possible to get 16 of
them together in one room and playing different parts in perfect sync.

In any practical sense, therefore, if you want to get 16 player pianos
to play together, it seems to me they have to talk MIDI.  The nice
thing about MIDI, of course, is that it doesn't care whether you're
using solenoid-driven, pneumatic, or purely electronic pianos -- and
therefore, you have the greatest choice available to you.  I would love
to hook up 16 MIDI-driven Ampico-equipped Boesendorfer Imperial Grands
and hear the Ballet Mecanique ... but I doubt that will happen in my
lifetime.

> The balance was totally wrong from the audience listening point
> of view.  Perhaps when transmitted as a broadcast recording it was
> better, since the pianos (each miked individually) can be 'turned up
> louder', and therefore be able to compete with the accompaniment of
> the drums, xylophones and other devices which clearly dominated the
> sound of the performance.

I actually agree with Douglas here; I wasn't that happy with the
balance, especially from the orchestra seats.  In the balcony, however,
it was much better, as it was visually, since you could clearly see all
16 pianos.  I think we may have over-compensated a bit in the stick
department: the percussionists were using the hardest mallets imagin-
able, in the expectation that otherwise they would be drowned out, but
this turned out not to be the ideal solution.

> Also, as expected, the Disklavier limitations meant that only partial
> keyboards were being used much of the time, often on doubled pianos.
> What had been the full sweep of a staccato chord on perforated rolls
> was now only part of the keyboard, and then the staccato wasn't that
> pronounced.

The chords were indeed split at some points, but not "over part of
the keyboard".  Instead, they were thinned out to accommodate the
Disklaviers' polyphony limits, with different thinnings used on
different pianos, but always keeping in mind that each piano covered
the full range, or very close to it, of the chord as written.

> When you could hear the pianos, together or singly, everything was
> pretty much at monotone level, and not very loud at that.  (One
> musical friend wondered aloud to me if 'overloading' the circuits
> might have made someone decide to cut the voltage, and this led to
> cocktail lounge playing on the part of the solenoid actions.  That's
> not my department, of course!)

The pianos were indeed all playing at the same velocity throughout,
except in certain passages where the velocities needed to be increased
slightly to accommodate particularly fast trills, a quirk of the
Disklaviers.  My article in Wired magazine goes into some detail over
our decision to not vary the velocities of the pianola parts.  I agree
(see above) that the instruments could have been louder, but this was
what we arrived at through much trial and testing.

As it happens, that hall is much, much louder when it is empty, but
we didn't know how big a difference the audience would make.  When we
did our recording session on Saturday in an empty hall, the difference
was dramatic.  Of course, the idea that some kind of voltage drop would
cause the pianos to play softer is absurd (it would have made the
computer fail long before that), but in point of fact in the middle
of the recording session on Saturday everything stopped dead during one
take: we had blown a circuit breaker.  It was *very* fortunate this
didn't happen during the performance!

> The Peress recording has mostly virtuoso keyboard pianists for its
> sense of life, and I heartily recommend it for those who wish to
> experience his well-researched recreation of the original orchestral
> version.

The Peress recording is indeed wonderful, but it really is a different
piece, with very different textures demanded from the live pianos.

> There will be, no doubt, many performances of George Antheil's
> experimental work in the future, and all will be different, as
> that's what the word 'experimental' means: a work-in-progress which
> was never completely finished.

Very well put!  That is indeed what the piece is, and it is what
allows us to have these discussions.

> One last thing here: the original movie, used in the Salon days
> of Ballet Mecanique, was 30 minutes long.  Censorship and artistic
> changes kept the movie in flux, and this is reflected on the score,
> where the composer keeps changing numbers for the 'time space'
> sequences. ... Originally, the Synchro-Cine film was three reels
> long, approximately a 1/2 hour in length.

I have found absolutely no other source that talks about the movie ever
being censored, or that it was ever more than about 12 minutes long.
And now you are saying it was actually 90 minutes long.  Perhaps you
could possibly direct me to one of your sources, as I would like to
investigate this further.

Thank you for the opportunity to respond on this forum, and all best
wishes,

Paul Lehrman