When was a "negative" clinical trial big enough? How many patients you needed depends on what you found.