Abstract
In food industries, chemical industries and others, many characteristics of products depend on the proportion of components of products. The design for studying the relation between the characteristic and the proportion is called "mixture experiments". Data in mixture experiments have the restriction that the sum of mixture components variables is 1. Then mixture experiments are not experiments with orthogonality. In this paper, the method for finding the property of main effects in the additive model with useful contrasts for estimable non-linear components is proposed.
