I hadn`t considered radiative cooling (the stefan-botlzman law!). What is your feeling for the relative contributions of convetive and radiative cooling in the Androcles reactor? I get the impression that radiative cooling must be only a small correction to convective cooling and so the knee would not be too visible.
if the reactor is free in the air, for air (convective) you have coefficients of the order 10-100 Power / Area / Kelvin probably 10.
for radiation you have W = sigma (T^4 - T0^4) = sigma (T^3+T^2T0+TT0^2+T0^3) (T - T0) => T=400K -> 10, T=500K 15, T=600K 23 Power/Area/Kelvin
So very well after 100 degrees celcius it can be the order of the convective cooling