Ooops... my intuition screwed up on this one: Since the effective specific heat does not remain constant with temperature -- there is a discontinuous rise at the boiling point -- there is a dramatic rise in the effective heat transport with temperature at the boiling point (whatever it is for the pressure in the reaction vessel).
That's all that's required for temperature control. Rossi's effort required to achieve self-sustained mode then would have been to ensure that the flow rate around the reactor vessel was in the range where it is low enough not to quench the reaction in the 1deg/gm/calorie regime, but high enough that in the regime where its some huge number the reactor reaches an equilibrium with the heat transport. I plead lack of sleep.
