I would just slam a 100TH tube in there or use two 100th tubes to balance things out a bit. They are listed on eBay all the time and there may be plenty in England left over from WW2 from the Lend Lease Program. The filament current is the same so I don't see why they should not work?
I agree, but first do a little troubleshooting. Is the 75TH bias off the Bias adjust rheostat steady, and pure DC? In the schematic the filter caps (C401, C402) are only 4uf + 4uf to make -145v bias to the Rheostat and the PA grid current meter. Increasing the PA grid current will boost the - bias to the modulator somewhat. There is a possibility of some interaction there if C401, C402 on deck D are not filtering, or the bleeder current on the bias supply from R401, R402, R403 (750 500, 1000 ohms respectively) is not about 64ma (145/2250). Making C401, C402 5X stock value might be a good thing.
As to the 75TH's, those metal plates like that are designed to get red hot at normal ratings. Might take 1 plate cap off, and key up (audio gain down). If OK, put the plate cap on, and then remove the other one. If either tube is gassy, there should be some glow within the envelope. This is a way to see if the idle currents are balanced, or if one tube goes into a runaway situation.
BTW, I've run 808's before at 2250V B+ as class B modulators. Takes about -50v of bias. These look similar to the 75TH, but unfortunately have a 7.5v filament voltage.
Could always get a 304TH and take it apart! There are four 75TH's inside.
Jim
Wd5JKO