[morzh]

I have found an abnormality and I think know where it comes from.

I did remember that there was a little arcing between the dial and the hand when tuning.
So, first I put a large (47uF) cap between the chassis and the last electrolytic decoupling the filaments.
The hum went away altogether.
There is a 0.1uF between the chassis and common.

So then I discharged the cap......not lot of spark.
Then I measured the voltage....and saw 120V. Except that the chassis was positive. Good thing I did not hold that cap there for a long time.
I reversed the cap, and this time after discharging there was a large healthy snap.

I measured the current via 10K resistor - just to make sure it is not capacitively coupled ghost voltage, but no - it was 10mA.
So the chassis voltage is almost equal to the rectified voltage.
But there is no measurable resistance between them; the voltage occurs at the same time as on the rectifier tube K pin.

So....I put two and two together, and I think my drop power resistor (the one mounted proud on top the chassis, riveted to it) has the leakage which develops when it is powered up. I desoldered the wire going to the tube and left the wire going to the filter, and replaced it with 82 Ohm 5W resistor. The hum got greatly reduced. (I cannot unsolder it altogether as it houses the second and third resistive sections of 970 Ohms). So I think I will replace the whole thing and see if it gets completely cured.

I thought of just using the electrolytic cap to remedy it as then it would work and most time it would employ the battery anyway....but it is not good to leave it like this.