Wiring up the bias supplies: Two negative voltages, around -4.5V and -35V, using an old trick from the past: back-biasing.
Instead of returning from chassis-ground to the transformer, you run through a resistor network to get negative voltage drop across the resistors. It's possible, because the main load is a two stage class-A amplifier, so there is a predictable average current running through it.
Fluctuations are taken care of by buffering the voltages with capacitors. Grids should not draw current in nominal operating mode, so you can use very high resistor values (like 500kOhm-1MOhm) for biasing. For stabilizing the voltage, run it through a 100k resistor to a capacitor.
Way back then, even small caps like 30uF were HUGE. That's why that resistor trick was VERY helpful, cause you can use an 1-2 uF cap and get same stability.

I am using big form factor capacitors on purpose. Those foil caps are available off the shelf and come close to what would be available back then regarding the form factor. You can get a 2-5uF cap for that voltage that's smaller than your fingernail, but that would have been science fiction in the 1920s, so I bought caps with same capacity but more higher voltage, as they are as chunky as the caps of same capacity way back then.
Why I don't use genuine parts? Those are pretty much all degraded and non functional by now. IF I get a functional one by chance, I'll replace the modern part :)

I am using modern resistors, as I don't know the final values yet. If the project proves worthy of beeing more original, I will replace them with 1920s parts. However, they are super rare and hard to find and quite expensive.
To make it more like early 20s, I could wind resistive wire onto a ceramic tube - will prolly do that :)