Dave Saul said, after suggesting that Welte used carbon microphone
techniques to measure keyboard dynamics for the Welte-Mignon:

> It might also explain why so many unlikely (and often technically
> unsound) schemes involving carbon rods, pointed wires, and troughs
> of mercury were perhaps "leaked" intentionally into the rumor
> mill to protect legally risky technology from all external scrutiny.

Be it known that one of the most successful early telephone
microphones, designed by a Welshman, Hughes, did not use carbon
granules but two carbon rods placed touching at right angles.  This
transmitter was known for its stability, reliability and freedom from
packing of the granules that the standard "button" type exhibited.

The Hughes telephone transmitter lost out because it was a lot less
sensitive than the carbon granule type, but it remained popular for
local phone systems for twenty years, i.e. 1880-1900, because of the
clarity and lack of hiss of its signal.

Hitting one with the underside of a piano key strikes me as a promising
method of getting a reliable readout signal.  Carbon granules would have
been useless.  It's possible the readout involved an electro-sensitive
paper roll with electrodes touching it: FF (fortissimo) = a bigger blob.

Dan Wilson, London (was in 'phones from 1945 to 1991)

 [ See, for example, the image of the DeLuxe "seismograph" recording
 [ at http://mmd.foxtail.com/Pictures/Welte/seismic.html  -- Robbie