The analysis of variance is not a mathematical theorem, but rather a convenient method of arranging the arithmetic.

Discussion to 'Statistics in agricultural research' by J.Wishart, Journal of the Royal Statistical Society, Supplement, 1, 26-61, 1934.