A study was made to determine the sample size for research on corn resistance to corn ear worm Heliothis zea (Boddie). The damage was measured using the Widstrom centimeter scale. The criteria used to select the sample sizes were such that a difference between two treatments of ten percent or twenty percent of the overall mean should be detected by the statistical test, and that each treatment mean can be estimated with a standard error of 0.10 y or a standard error of 0.20 y. The data were obtained from two field experiments, carried out in Instituto Agronômico, Campinas, State of São Paulo, Brazil. The treatments were: cultivar Asteca Prolífico VRPE VII, which is resistant to Heliothis zea; Hmd 7974, a susceptible double hybrid; cultivars Maya XVI and HS 7777. A randomized complete block design with ten replicates was used, and each plot consisted of four rows with 50 plants per row. The variance components method was used to define the type of sampling. The most important variance component was due to the variability among units within rows. The following minimum sample sizes can be used to detect the difference between two means: 6 blocks with 4 rows of 24 plants for a difference of 0.10 y and 5 blocks with 3 rows of 6 plants for a difference of 0.20 y. For estimating the mean damage with 0.10 y as the half confidence interval on the mean of the population, the following minimum sample sizes can be used: for cultivar Asteca, 7 blocks with 4 rows of 30 plants per row; for the hybrids Hmd 7974, HS 7777 and cultivar Maya, 5 blocks with 2 rows of 24 plants per row. If a 0.20 y is specified, as the half confidence interval on the mean, it can be used 5 blocks with 3 rows of 12 plants for cultivar Asteca and 5 blocks with 2 rows of 6 plants for the other three cultivars.
sample size; variance components; population parameters; resistant cultivars; Wildstrom centimeter scale