Hand capacity was a side effect of poor isolation of the detector from things like
ground and antenna changes. I'd found that the same circuit with a down converter
ahead of it tended to be very stable and the small added gain of the converter allowed
coupling to the detector be looser which improved the selectivity.
It was my last shot at regens back then as I was both trying to be solid state,
small and low power and incorporate superhet design ideas for things like
stability and selectivity. Much of the early work was done using a regen in
the 400-500khz region where selectivity was excellent and far easier to
build then a full out 455khz IF.
Look elsewhere on the net for Regenodyne for someone that more recently started
replaying that idea. [see
http://www.qsl.net/w/wd4nka//TEXTS/REGENf~1.HTM]
While his is hollow state it could easily be translated to transistor or FET.
More modern designs of regens that are popular is the "Kitchen Regen" and the
"Desert Ratt 2".
Allison