I probably first saw the number from Fitch Williams. Likewise
the old saw that the 3600 rpm motors don't make good converters.

I looked at some of Fitch's old posts. He contents that any good
motor will make a good RPC. The original theory was that a 3600
RPM motor will have more mass and hold more energy for that peak.
Fitch debunks that theory with his own theory. His theory is that
the idler motor is a rotary transformer not a flywheel. If the motor
is big enough and has enough copper and iron to make the transformation
the motor RPM doesn't matter.

I have been following this thread and experimenting a bit on my own.
I now believe that a RPC isn't doing much until the load motor has
a significant load. Both motors are pretty much running on single
phase. At some point, the load motor starts to slow down(ie slip)
and the RPC motor is still running at full speed and starts suppling
some power to the load motor on the third leg.
Maybe my theory is all wet, but I have tried connecting up three
different motors without balancing caps and I see almost no current
in the third leg when the motors are unloaded.

chuck