We were together learning how to use the analysis of variance, and perhapsit is worth while stating an impression that I have formed-that the analy-sis of variance, which may perhaps be called a statistical method, becausethat term is a very ambiguous one — is not a mathematical theorem, but rather a convenient method of arranging the arithmetic. Just as in arith-metical textbooks — if we can recall their contents — we were given rules for arranging how to find the greatest common measure, and how to work out a sum in practice, and were drilled in the arrangement and order in which we were to put the figures down, so with the analysis of variance; its one claim to attention lies in its convenience