My colleague, John Lee, and I have developed a package that may be
interesting to amateur seismologists.  A paper has been posted at
http://physics.mercer.edu/hpage/mouse-sensor.pdf
   As noted in this article, John will make available free to anybody
who sends him an email request for a password--the LabView executable to
operate the mouse.
    The resolution, at about 50 microns, is too small for direct
application to seismology; however the method might be useful for
calibrating instruments.  Also, anybody wanting to monitor the earth
(stable pendulum as a plumb-bob) should be able to use this technique to
watch for changes greater than 50 microradians.  The typical diurnal
thermoelastic variation is about one-third that value, but over the
course of weeks to months I predict there will be observable changes
that are not being routinely measured in nearly the 'global' sense they
ought to be.
     To allow easier use of the sensor I removed some materal from the
bottom of my mouse using sandpaper.  By this means the gap-space over
which the unit can operate is approximately doubled.  Without doing so,
one has to be more careful with alignment and stability of the
pendulum.  With my present capability (as wide as 3 mm) it is very easy
to work with this instrument.
      Additionally, I have an idea that some of you may want to
consider.
    Any truly outstanding vertical seismograph requires force-feedback.
The traditional approach is one that uses force balance.  The same
'strong' force that keeps the instrument from even moving substantially
(much less 'goiing to the rails' because of temperature/pressure
changes) is also used to provide the required near-critical damping.
For more than a year I have demonstrated with my modified Sprengnether,
that a 'soft' force feedback has advantages over the traditional
approach.  The feedback I have used with my Sprengnether derives from a
long time constant integrated output from the capacitive sensor (large
dynamic range area-varying, fully-differential unit).  Based on my
experience with this instrument, I propose the following:
       Since the sensor used for force feedback can have terrible
sensitivity compared to the sensor used for detecting earthquakes--build
a `centering' (feedback) package that uses the mouse sensor.  The
requirement for DAC output for the coil of the actuator can be managed
(it appears) with a 'dirt cheap' homebuilt piece of electronics (an
example at
http://www.allaboutcircuits.com/vol_4/chpt_13/3.html  )
    Without a parallel port, your USB computer would need a printer
cable (< $20) to operate this R/2R DAC.

Randall