Say you are testing hundreds of hypotheses, each with a t-test. What considerations would you take into account when doing this?