Referring to Jurgen Goering's comments about the depth that Welte's
carbon rod plunged into the mercury filled trough.  As long as we're
having fun speculating on how it may have been done, here's my two
cents:

It's often speculated that the carbon rod must have gone deeper into
the mercury trough on loud notes hence creating a lower resistance,
more current, and a wider line impressed on the recording paper by the
inking wheels.  I don't think this can happen as a piano key stoke is
essentially the same whether a loud or soft note is played.

It seems to me for the Welte recording system to be able to measure
dynamics, that a voltage would have be generated that was proportion
to the speed that the rod plunged into the mercury.

If we assume that the rods were partially immersed in the mercury at
all times, an easy and direct method of getting a voltage proportional
to the loudness that a note was played is to apply a DC voltage to the
recording device and put an inductance in series with the carbon rod.

Without an inductance in series with the rod, the current would
increase as the resistance of the rod dropped as it descended further
into the mercury reaching a maximum at the instant the rod reached the
bottom of its stroke with the key fully depressed.

An inductance in series with the carbon rod, however, would resist the
change of current as the resistance of the moving carbon rod dropped
causing a voltage to appear across the coil directly proportional to
the rate of change of the moving carbon rod's resistance.  The rate of
change of the moving carbon rod's resistance, of course, is directly
proportional to how loud the note that it's attached to is being
played.

So here is a simple and direct method of producing a voltage that is
directly proportional to the loudness of the recorded note.  This
voltage then would be the voltage that pressed the inking wheel against
the recording paper.  There would have to be some probably touchy
adjustments and regulation to have the wheels so they were just on the
verge of touching the paper while their note was not being played.

This is pure speculation on my part and I have no idea if Welte may
have utilized this principle.  But this is a method that could produce
a voltage directly proportional to loudness, and it was well within the
technology of the day.

Regards,

Dick Merchant
Carlsbad, NM