HOW "old"? It matters.
> 1m5
If you are looking at the plan AErige linked, that is "1M5".
(Yes, capitalization *may* matter*.)
On that plan, all seems clear.
Cathode caps are 20 micro-Farad, because that bypasses 2K5 down to 3.3Hz; no other point-shift (F, nF, pF) makes any sense. And if you only find 22uFd, that's cool, nobody can prove response extends to 3.0Hz.
Plate-to-OT caps could rationally be 1.5 micro-Farad. We don't know the OT impedance, but it's a "25K" tube, a 100K resistor, an OT in the 10K-50K range is suitable. 1.5uFd with just 10K would be 11Hz. But consider the tube it's nearer 3.2Hz.
Now the funny ones. There's "100" from plate to ground. Taking total plate node impedance as 25K: If 100 Farad you'd need a truck. 100uFd would short-out ALL signal (above 0.05Hz). 100pFd gives around 60KHz: passes all audio, diverts radio.
It is, IMHO, very odd to "assume" pFd when no multiplier is visible. However this IS how many modern caps are marked: "104" is a "10" with 4 zeros, or 100,000pFd (100nFd).
Then the "4K7" might possibly be 4,700 pFd. Check: it must swamp the ~~30pFd of the capsule, check. It should hum-reduce the voltage coming in through the 180Meg resistor. 4700p against 180Meg is 0.2Hz, so all audio is cut 100:1 and line-freq even more, check.
