Flare Dammit - I don't know why it does it either - that's why I posted the question to Nick.
Helmet Fire - I completely agree with the business about torque spikes and the overall change in rotor drag (see my post on the Aerodynamics thread). However, in searching for an explanation for the disparity in roll rates I have tried to consider all sorts of different ideas. It is a fact that the slope of most blades CL curve is not constant - generally at low AoA the slope is shallower, increasing in steepness towards CL max. My suggestion therefore, was that as the retreating side is operating at higher AoA than the advancing side, then the increase of CL on the retreating side during the roll was greater than the reduction in CL on the advancing side.
I will be very happy if Nick can come up with an explanation.