> what should be the power rating of that 2M pot?
The speaker load for two 6V6 will be about 10K at the OT primary.
So a "2 Meg" seems like it would do nothing at all.
Such a plan was common on table radios. Usually a 5K pot across a 2K load. At full 5K, little effect. At half-way, 2.5K, output is reduced. At "1", about 500 ohms, output is very reduced. Only in treble; the cap blocks bass.
IIRC for a 1W 2K radio the cap was often 0.05uFd. Taking the 2K impedance, 2K against 0.05uFd is 1.6KHz. That's a bit brutal top-cut, but for AM radio or 78s you might need brutal.
If your signal had LARGE power above 1.6KHz, the pot must absorb the *entire* 1 Watt.
This 'could' happen in testing.
But speech/music systems like radios (and guitar amps), playing for people's ears, generally will NOT have huge power above 1KHz. A tweeter crossed-over at 800Hz takes less than a third of the total power; at 2KHz, less than a tenth. On those grounds, the 1 Watt radio could get away with a 0.1W pot. Since cheap pots are 1/4W, it's not a problem.
Now the model 26. That's about a 10 Watt. The pot should be rated at least 1 Watt. Perhaps more for modern bright/fuzzed guitar sounds. And the resistance should be somewhere above 10K, so full-up is full-bright, but part-up is part bright. 20K or 25K makes a lot more sense than 2 Meg.
I think that's why we don't see this type tone control on anything bigger than a Kay or Kent (cheap junk derived from table-radio parts). Why use a costly several-Watt pot in the output stage when a cheap pot in the preamp will work? (Guitar amp has more audio preamp than a radio, therefore more places to hang your tone control.)
> if that conjunctive cap failed by going short
Then the pot burns up. Small smell, no big deal.
If the pot happened to be turned to dead-zero ohms, the sound would stop. For a while. I suspect any reasonable power pot on a two-6V6 amp would burn-up even when turned to "zero" ohms (there's always a little parasitic resistance under the wiper).