If you're serious about it, run a bunch of experiments, graph the results, then find a curve fit.

Reminds me of a big science experiment we had to do in the last year of highschool. We were given a week off to work on a large experiment, any experiment we wanted. So the two of us chose the Mpemba-effect, which claims that warmer water freezes faster than colder water. Counter-intuitive, so it's fun to test empirically. Your post made me dug it out again.

In retrospect, we were quite thorough about it, testing for lots of possible factors: the position in the freezer, demi vs normal water, hard vs not hard water, shape of the glass and the volume of water. No effect. Only when we added ions (calcium, magnesium) we managed to delay the freezing. The only way we managed to get an effect is when we boiled water with ions (to get rid of the ions) and froze it simultaneously with water that still had a lot of ions in it.

The best part of this all was that it was doable in a day or three, in the freezer in my home, while the other groups had experiments that lasted the full workweek in the chemlab. One group tried to make an e-ink screen with ferromagnetic fluid and magnets but failed when the chemistry teacher accidentaly destroyed their setup.

I'm not really sure why I tell you this. I hope you didn't mind.