[Arran]

Arran, if you look at the schematic, you will see that the tubes are biased by a resistor on the center tap. One could set the circuit up with the filaments in series and it would work OK. Besides the filament voltage, the other tube specs are really fairly close between the 71A and the 45. http://www.nostalgiaair.org/pagesbymodel...029582.pdf. I only suggest it because it is the only plausible reason to have 2 45s in the push pull section. As I said, I think someone just stuffed them in to fill it up.

 The simplest explanation is usually the correct one, that this was just someone grabbing some random four pin tube to fill some holes. I don't know where Kirk got this set but this is quite a common practice just to make a set look good in a photo.
  The bias resistor for the two #71As (which is one section of part #25) is in series between the center tap on the 5 volt winding and the center tap of the power transformer. I'm not certain about how well that arrangement would work connected across two #45s is series, one minor point is that it would be the wrong value, but more importantly the reason that they are connected to a center point across a tube filament is that they are also used for hum ballancing the two ends of the tube filament. In some radios they use a center tapped resistor across each tube filament, in others they use a an adjustable wire wound resistor, but I've never seen any radio that used two directly heated AC power output tubes with the filaments connected in series. Now if the total load of the #45s was only .5 amps, I'm sure that one could rig up a way around the existing bias network, but the two #45s will still draw a current load of 1.5 amps across a winding designed to accept 1/3 that much at the same voltage.
Regards
Arran