"Go too fast and the turn rate reduces, go too slow and you stall before you can pull the required g to get a higher rate of turn. So why are they finding this is happening at a lower power setting? Shouldn't they be flying the turn at max power and using the wing loading to keep the speed back? I think what is happening is that if they genuinely flew max power max g, they'd be pulling more g's than they are comfortable with."
This is exactly the question I have been asking myself: However these aircrafts could not sustain more than a very reasonable 3.2-3.4 Gs in continuous sustained speed turns, so this was well below what would be unsustainable for the pilot, especially in a combat situation: Given the vital need to gain on the opponent, why would less power be used?
Even a broader circle, if it is completed faster, is still a gain in sustained turn rate, and thus a gain towards an opponent's tail.
There is other corroborating evidence to the existence of this puzzle: One Me-109G-6 ace mentionned that the best sustained turn speed on his machine was an incredibly low 160 mph, barely 55 mph above stall, specifically mentionning reducing the throttle (and opposing that to what most other wartime pilots were doing).
When giving out full power (for a maximum of around 400 mph straight), this German aircraft type cannot turn hard enough continuously lower its sustained turn speed much below 200-220 mph, which is a good indication of how much the throttle was reduced by the Finnish ace to achieve what he considered the "best" sustained turn speed of 160 mph.
The problem with these types of aircrafts is that they were always short on power, so more power in turns for these prop types should not be a bad thing that would put the sustained turn rate outside of the pilot's indefinite endurance: 3.2 Gs is barely half the maximum the pilot could tolerate in unsustained high speed turns in those machines...
If that was so, why reduce the throttle, and maintain it there, when speed is already low?
Furthermore, even making a smaller circle at a lower speed will cause the same exact disconfort in applied Gs if the same turn rate is to be maintained: Turn rate and Gs are correlated it seems to me: Except for slight gain in gunsight lead from being "inside" in a smaller circle, there is no advantage in G confort to reducing the speed and still be completing the smaller circle in the same amount of time as a broader but faster-speed circle.
Since these WWII aircrafts lacked the power to black-out the pilot in sustained level turns (surrendering altitude in turns being generally a tactical no-no when "locked" in sustained low-speed turns, or a moot point when near the ground anyway), the only reason to reduce the throttle and find an advantage is if the smaller circles at a slower speed (and thus at a lower power level), were actually completed faster, because they were so much smaller in diameter compared to the loss of speed.
But reducing the power alone to achieve this doesn't make sense, as one would assume the reduction in circle diameter would be proportional to the reduction in speed, nullifying any circle completion rate advantage or worse, especially when speeds get as low as the above-mentionned 160 mph. (At a very low near-ground altitude, so no huge IAS-TAS discrepancy)
The only way to produce an advantage in sustained turn rate at a lower power level, it seems to me, is if the lower power actually reduced the "real" in-flight wingload in some way: This would allow making the circle disproportionately smaller compared to the loss of speed of the reduction of power, thus gaining a sustainable advantage (where the different thrust location of jets might not cause the same effect, hence the absence of this tactic in the jet age)...
Hence my interest in finding out if any variations in wing bending vs power output in level turns, for nose-driven aircraft types, has ever been observed and measured, or if any wing-bending stress gauge tests of this kind has been done on similar-configuration aircrafts.