RT Journal Article SR Electronic T1 Use of Pretreatment Information in Design of Experiments. I. Nearly Optimum Designs for Covariance Models JF American Journal of Enology and Viticulture JO Am J Enol Vitic. FD American Society for Enology and Viticulture SP 174 OP 177 DO 10.5344/ajev.1980.31.2.174 VO 31 IS 2 A1 S. Geng A1 M. B. Carter A1 J. O. Johnson A1 S. K. Perng YR 1980 UL http://www.ajevonline.org/content/31/2/174.abstract AB Pretreatment information about experimental units is often available for agricultural experiments. This information has been used primarily for adjusting treatment response at the stage of data analysis, whereas its use to improve the experiment is seldom discussed. If a covariance model can be established to represent the relationship between data collected before and after the treatment, a procedure is proposed for construction of nearly optimum designs to minimize the generalized variances of treatment means. The procedure is illustrated by allocating grapevines (Vitis vinifera L.) to two and four treatments based on pretreatment measurements of pruning dry weight and volume of aboveground structure of plants. It is shown that the group means are well balanced after the directed allocation.