This subject comes up again for me as i try to order my old emails.
Sorry, i may have missed something, but in Jim's article, what does this mean, for coupling cap, like "5+3 pF".
Surely it does not mean, really, 8 pF ?
Unfortunately my RDH book is hiding from me right now, but offhand i don't understand how you can calculate
selectivity from the coupling factor without knowing Q.  Looking at the graph from the ARR-15 manual, i have to
guesstimate overall Q as 500 kHz / 7 kHz, my crude eyeballing of the not highly defined graph seeing -3 db at
about -7 kHz. The Q is affected by coupling, of course, but also the Rp of the tube plate ( ~ 250 - 350 k ohm ?? )
and the following tube grid. Those are probably not decisive factors compared to the actual IF coil Q.

After seeing Dave Stinson's impressive  vids of the BC-229 receiver with the adjustable feedback, i am giving that
kink a rethink.  I tried that kink decades ago in a Command Set receiver but discarded it as near worthless.
I cannot recall if i disabled the AGC and i'm thinking i did not. So maybe if that is done, this idea is actually worthwhile ?
On the other hand, i recall trying it in such receivers as the Heathkit GR-64 and so on, where they had a regenerative
IF by lifting grid 3 above ground via a pot, and i was not at all impressed with its working.
I recall that Roy Paffenburg booklet on the ARC-2 used the regen IF idea to get more selectivity.