I agree completely with Albert, if you look at the nominal voltages on the schematic the output transistors of U4413 should have a collector current of 36mA ( (50-37)/360R ) so with any reasonable current gain they should not draw more than a couple of mA of base current. Albert measures 4mA for the total base current into both output transistors which is eminently resonable.
Neglecting the base current into Q4420, its base should sit at 14.88V ( (5k / 16.8k) * 50V) and its emitter should be around 0.6V-0.7V more positive so 15.5V - 15.6V and Albert measures 15.45V. The function of Q20 is to provide a low source impedance to bias the output transistors of U4413 at 15.45V rather than the 15V whichTek could have obtained by connecting R33 directly to +15V. This also implies that R4419 should be a low inductance type to maintain the frequency response of the amplifier.
I think it is safe to say that any junction transistor which measures a forward Vbe of 2.6V is no longer a transistor! If you take Q4420 to be open circuit from base to emitter then R4420 and R4419 supply the base current to the output transistors without any voltage regulation which is why the bias voltage shoots up to 17V.