That's interesting. Personally, I think it's easier to maintain same power in a headwind than a tailwind. I always presumed it was a combination of differences in crank inertial load and in convective cooling.