| Abstract: | This paper calls attention to an experimental design which enhances the efficiency of the technique described by B. F. Houston and R. A. Huffman 3]. This alternative design is a modification of the simplex lattice design developed by Scheffé 6] for the mixture problem. While not possessing the optimal statistical properties of the central composite design proposed in 3], such as minimum bias and minimum variance, this modified design is optimum with respect to the number of design points employed; hence, it is the minimum cost design. Further, the design proposed in this note is excellent for use in multiple-block experimentation, which is very often required with constrained systems. |
