Two corn experiments were carried out in Santa Maria, RS, Brazil with two factors (cultivar x density) in a random block design with six replications during 2000/2001 grown season. Thirty medium and 16 early corn cultivars were used in two plant densities. The objective was to find phenological and morphological traits that increase experimental error of grain yield in corn. Average and variance of grain yield, days until male flowering, plant high, first cob high, plant and cob number were estimated for each cultivar and plant density. Linear correlation among traits was estimated for each experiment. With the correlation matrix a path analysis was done to verify direct and indirect effect of each trait to grain yield variance. Medium cultivars of corn with uniform cob number per plot show to have a reduced grain yield experimental error.
experimental precision; path analysis; modeling; Zea mays
