L. P. Reitz has very simply stated one of the main corollaries of plant breeding, i.e., "Varieties fall into three categories: those with uniform superiority over all environments; those relatively better in poor environments; and those relatively better in favoured environments". While plant growth models carry a huge potential for the future of plant breeding, the present state of knowledge leaves the breeder with a highly empirical system of operation that is being used very widely and almost exclusively. The analysis of variance may be performed according to two models. When varieties, locations, and years included in the experimental sample can be regarded as a random sample of all possible varieties, locations, and years, the model fitted to the data is known as the random-effects model. In the analysis of variance, a sample of genotypes and environments could be evaluated as a series of orthogonal comparisons with the appropriate number of degrees of freedom.