This user did a lot of work, but only with one side of the circuit.
http://www.youtube.com/watch?v=DzqCnIKP12U (http://www.youtube.com/watch?v=DzqCnIKP12U)
This is a big file (take a while to open the link)
http://www.rexresearch.com/hendershot2/HendershotMotorMystery.pdf (http://www.rexresearch.com/hendershot2/HendershotMotorMystery.pdf)
http://www.scribd.com/doc/95105/The-Hendershot-Motor-Mystery (http://www.scribd.com/doc/95105/The-Hendershot-Motor-Mystery) scribd version
Basically there are two loop antenna (thought of Lidmotor and his am radio with loop antenna). The coils are a basket weave, which decreases capacitance, and provides high Q by having a seperation of the windings the same as the wire size basically. They are wound very wide (6 inches). A 40uf foil cap is dismantled, and the 91 inches of 2 3/4 inch foil is re-wound on a 5 3/4 steel (ferrous, no mention of copper/aluminum/etc, a theory of operation is that it is stimulated by the rotation of the dynamo that is the earth's core, so it should be magnetic) . Then there are a series of 3 other oscillator circutes directly related to one loop antenna, and then the same 4 replicated. Then also two 1:5 transformers are used, asymetrically. The buzzer should oscillate between the two, alternating generating voltage spikes as the flux makes/breaks contact. Should think that the solonoids for the buzzer could be made solid state with something like stacked pnp's(?). But then would I use coils there where the solonoid would have been (with a core?) or replace with a capacitor?
Since they are probably in slight distune, say by as much as 50-60hz, they generate a voltage potential? Could use some ideas with replacment parts for the solenoid part of this...
Sorry that was a horrible place to start, although interesting
http://www.youtube.com/watch?v=rfe0zGQ2aNg&feature=related (http://www.youtube.com/watch?v=rfe0zGQ2aNg&feature=related)
Here he's testing his coil with radio stimulation. but that's not the point, he didn't have two :)