[Arran]

I would go with the measurement points suggested in the service data. Many of these 1930s sets don't really have a common negative to everything because of the bias networks they have set up so they use voltage measurements with respect to cathode or "K" as a reference point which is as close as one can get to common negative. The plate will always be positive with respect to the cathode, same with a screen grid, the control grid will usually be negative with respect to cathode. It's hard to say how much the original voltmeter would have loaded the measurements down, if it was a standard analog voltmeter it would have, if it was measured using a VTVM then it would not have. I would just measure the values of the carbon resistors and forget the voltage chart, if any are more then 25% out change them, if any voltages are way out then those would be the cause anyhow.
Regards
Arran