Results are very close. I think I'll also do 2 of 5 tests between each pair of roasts tomorrow and see if there's a legitimate tastable difference going on. If there isn't, it's harder to make the case that the variation matters.