While we have not yet arrived at a confirmation as to how Welte worked
this system out, at the same time no one has come up with a reasonable
explanation of how they may have done it with the technology of the
day.  So I want to just have a little fun with this and ask anyone
interested to throw in their two cents too.  It's fun to speculate.
Maybe my initial idea will spark a really good one!

The way things are done in the technical world is basically in secret.
The last thing an industry would ever dream about doing is to patent
an in-house design that gives them an edge in the market and cuts their
workload.  So patents of the day represent technology understood and
utilized usually 10 to 70 years earlier.  I think that about anybody
will agree with that statement as a general rule of thumb.

I say this at the outset, so that we don't have objections based on the
patents of the day to throw cold water on what they could actually do.
Remember, Tesla presented the math and theory for radar in 1917, but
had worked it out before 1900.  The only thing that kept him from
building one was money, his workload in other fields, lack of competent
help, plus the lack of certain materials and industrial processes
required.  Otherwise, it was all doped out.

By the time of WW1 there had been many great advances in the field of
motors, transformers, and general magnetics.  Practically all the basic
math had been done and the devices just poured out the doors of
research labs across the world.  It was common knowledge how to build
magnetic amplifiers using control windings which saturated or blocked a
magnetic flux circuit.  Decoupling power coils with a tiny little
control winding was simple.  So decoupling them by any given percentage
would also be simple to do.  You just decrease the efficiency with a
counter winding.

Magnetic fluids and inks were not hard to make either.  The most
expensive thing would be the master paper rolls, I would think.  I
don't know how they justified that expense.  But as far as planting
conductive paths of differing resistance on the master rolls (or a
double track for each resistance path, easier to control), I see no
problem whatever.

There is no way to know details, but I do think it would be a good
exercise to establish several things here:

1.  Did they have the technology to do it?

2.  Would they benefit from the technology economically and musically?

I think the answer to both questions is yes.

Consider now just one possible note circuit.  The carbon rods are not
actually connected to the keys, but are spring-loaded and mounted in
guide tubes, aligned under the keys.  The thing attached to the key is
an adjustable plunger that hits the rod down.

The rod has a mercury-wetted contact on each of its ends, assuring good
connection, and is floating on mercury in its little separate tube (you
don't need a hog trough containing a couple hundred pounds of mercury
to do this).  The key plunger is adjustable, depth-wise.  It is also
metal, and wired to the other end of the key, where a second long
spring wire connection is made in series with the damper lever.  In
this key circuit is also contained a potentiometer to zero out the
control current of each note circuit.  When a note is depressed, the
ink pen comes down but no dynamics are recorded this way.  Only if the
rods are driven down into the mercury to partially short the total
resistance does the threshold signal something other than a brushed
note.

Regarding brushed notes: with this hookup and a little distance
adjusted at the damper levers, they would be eliminated.  Of course,
in the high treble, you would just extend the contacts without any
dampers.

So we now have one wire for each note and a common wire for all notes,
which leave the piano and operate the master recorder.  The circuit was
probably DC at this point.  Whether each note had its own dynamic
sensor at the recorder or not isn't important, because the circuits
could just as easily been divided at that point into bass and treble,
the higher voltage signal at any given moment being predominant, and
translating into the Welte format, at least to the degree that it would
be then easy to translate from this record into the finished punching
master roll.

Let's say then that the master roll uses dynamic information stored
along its margins and the system is divided into bass and treble.  That
would decrease the number of proportional inkers considerably, and they
could be easily used as tracks to just continually adjust the vacuum of
the piano in playback, or directly as a transformer/solenoid system
doing the work.  From there, it really shouldn't matter if your piano
is vacuum or electric.  The question is, how would Welte get from the
electric master to the piano control?

Here is one theory: The master was coated in a sort of "developing
machine" that gave its inks the necessary conductivity, because
conductive ink is very heavy-bodied and could not be deposited directly
by the pens.  That machine might be a converted web printing press
whose "inkers" had good control of the graphite and drier which stuck
to the ink.  The excess graphite was mostly removed from the non-
absorbing, slick finished paper with spinning brushes.  The master was
then dried on its way to being respooled.  That would not take very
much time.

The resultant electrical master would have marks on it that you could
actually feel with your finger.  They would be considerably raised, and
hardened by heat.  If you burnished them, they would look like little
metal contacts -- which actually, they were -- and they would be as
precision as regular printing.  The dynamic marks' resistance would be
proportional to their width.  If the contacts which read these marks
were separated with an insulator of precisely controlled thickness,
sprung against the paper as it rolled over a platen, and if the marks
were a bit longer than the note played, due to the diameter of the
marking pens, then the contact pair would likewise be the same width
(going across the marks) to subtract out that extra length.  Now we
have a one-to-one master roll whose errors are beautifully canceled out
geometrically, and we have something that can actually control a magnet
circuit, as long as very little current is used.

Let's suppose now that the resistance on the roll paper becomes a
shorting resistor to a control winding of a transformer in an ac
circuit.  And let's also suppose that the normal current through this
control winding blocks energy transferred to the secondary, normally.
It would stand to reason then, that were you to momentarily short the
control winding, you would actuate whatever it was that the transformer
would control.  And to the degree that the resistor across the control
winding was small or large, would be the degree of energy allowed
through the flux circuit to the secondary.

From here, it's a walk.  Now I don't know much about deposited
resistance, but I feel as though the technology was known, since they
were playing around with carbon mikes and that principle, long before
the advent of wireless radio.  And what we learn later on, which came
out in print, was long after the fact, and after many engineers had
been playing around with it for quite a while.  For example, if mech-
anical TV had its public debut in 1925, then how long had they been
researching it and perfecting it in the lab before they unveiled it?
I think you'd have to admit, at least 6 years.  You might take that
long just to get the funding to produce the idea in the lab.

I'd sure like to hear some engineer's ideas about this, too.  I'm sure
I have missed a lot, as far as possibilities of the time were concerned.
But I am also sure that the idea was not just possible one way, but in
many ways, and that the technology of the day was easily up to the task.
It was the question, 'Is it economically feasible,"  rather than, 
"Can we do it."  Of course they could do it.

Craig Brougher

 [ The magnetic amplifier was a compact alternative to high-power
 [ vacuum tubes for low-bandwidth needs.  WW2 battleships used a 'mag
 [ amp' weighing maybe 70 pounds to amplify the targeting commands from
 [ the ships radar going to the electric motor system which steered the
 [ big guns.  More than one seaman was injured by a swinging 5-inch
 [ gun barrel -- it *moved*!  The mag amp was used for many years until
 [ power transistors took over the task.  -- Robbie