Copycat - The method I'm planning to use would be to separate rowers into groups of 3-4, then give each group only one supplement, including possibly a placebo (Vitamin C, Sugar Pill, or Fish Oil - whatever I can find that looks like a supplement). The groups definitely wouldn't know exactly what they're taking. If I wanted to make it double blind, I could make it so anyone administering the products also didn't know what was what - but "double-blind" just seems like overkill for this.
Results would be compared to each individual rower's recent (1-2 weeks prior) performance during an indoor rowing machine competition.
Not entirely ideal, but I'm heavily constrained by sample size and PITA factor.