> The "main" o/p xfmr's low side is tied to chassis.
The whole thing frightens me. Chassis-to-line isolation seems hit-and-miss, ad-hoc as a simple mono recordplayer grew stereo and reverb. Also the caps are surely lowest-bidder, and now 40 years old. Also, I've found a 2268A schematic showing alternate grounding. And I know someone who worked in a similar factory... mistakes happened.
> I am going to add an isolation xfmr, 50va or 75va.
Essential.
Power seems to be 19 Watts in heaters and 36 Watts worth of DC. The 50VA may work; 75VA seems a much safer bet.
Tie Green wall-wire to chassis, solid, so it CAN'T lose contact.
Then feed the 120V _secondary_ of the isolation tranny to the original line terminals.
Then add chassis-mount _metal_ guitar jack, and bond the sleeve to B- near the small tubes.
VERIFY dead-zero volts from chassis to wall-plate screw, amp on or off.
With this connection, none of the dubious "isolation" caps matter. (They should be shorted-out by the chassis to B- bond.)
You "could" jumper-together the two top-lugs of the Volume pot and feed guitar in here. Then jumper together the speaker leads from the OTs. Since each was 8 ohms, the parallel connection is 4 ohms. (You could clear all the other stuff from the second OT speaker leads, and wire both windings series for 16 ohms.)
Input sensitivity is about 0.5 volts. It will take a hot pickup and a strong arm (or a booster pedal) to coax all 9 Watts out.
That's if it even plays. Junk like this, almost all caps are bad or will go bad soon after awaking from a 30 year sleep. Being very low-bid parts, even resistors may go bad.
How ugly is the construction?
I'm thinking to tie-together both 7695 and OTs, treat that like a Champ's 6V6 and OT, then (reluctantly) torching everything else off and building a Champ-like front-end.