Yates's method for testing factor main effects in two-factor models has been extended to general linear models. The sum of squares he introduced is now known to be equivalent to the difference in error sum of squares between restricted and full models, making it a unique and efficient way to test specific hypotheses in these models.