It depends how big the effect size is. You can conclude that falling off a 100' cliff is harmful with as little as n=2. Here, the effect size is big enough to g

It depends how big the effect size is. You can conclude that falling off a 100' cliff is harmful with as little as n=2. Here, the effect size is big enough to get significance with n=21. For tiny effects like in priming studies, you need much larger samples.