A fresh look at the basic principles of the design and analysis of experiments

When Professor Neyman invited me to attend the Fifth Berkeley Symposium, and give a paper on the basic principles of the design and analysis of experiments, I was a little hesitant. I felt certain that all those here must be thoroughly conversant with these basic principles, and that to mull over them again would be of little interest. This, however, is the first symposium to be held since Sir Ronald Fisher's death, and it does therefore seem apposite that a paper discussing some aspect of his work should be given. If so, what could be better than the design and analysis of experiments, which in its modern form he created? I do not propose today to give a history of the development of the subject. This I did in a paper presented in 1963 to the Seventh International Biometrics Congress [14]. Instead I want to take a fresh look at the logical principles Fisher laid down, and the action that flows from them; also briefly to consider certain modern trends, and see how far they are really of value.