[chuckm]

I'm not so sure, but I would not be surprised if the 4sqin per 25 hP is accurate for anything 35mph and higher. Here's my thinking:
1) The air permitted through a radiator does not scale linearly with speed. Though I don't have anything to support this other than a intuition and a Scanguage, I suspect that most radiators produce the highest ratio of cooling to engine heat production around 40mph.
2) Engines can, depending on the engine bay configuration, can shed an awful lot of heat even without the aid of a radiator, hence air-cooled small hP engines. When you are operating your engine at low load, the actual cooling requirement is relatively low, but still higher than what could be taken care of through natural (non-forced) convection, especially with a closed engine compartment. Stop-and-go compounds the issue because of the intermittent medium/high loading. But drive a constant 25mph and your fan probably will never come on.
Jack, for your purposes, I would still probably end up around 16-24 sqin of radiator opening. That's plenty of safety factor to compensate for not having a team of aerodynamicists to herd the right amount of air where you want it, not having perfectly optimal radiator placement, and a host of other unknown factors. I'll sacrifice a point of cD to insure I don't burn an engine up. The penalty for an opening that may be 6" x 4" can't be that astronomical. The penalty for a burnt engine will be much higher than any gas you'd save.