The aim of this research was to determine how many evaluations (replications) were necessary to predict the performance of common bean cultivars. Fourteen common beans cultivars (Phaseolus vulgaris L.) were evaluated in nine experiments conducted at Santa Maria, in Rio Grande do Sul State, Brazil in the agricultural years of 2000/2001 and 2004/2005. In each experiment, it was used a randomized blocks design with three blocks or replications. In each block d 14 cultivars were evaluated. Thus, the causes of variation were blocks, cultivars and interaction blocks x cultivars which is the experimental error. The following characters were measured: number of pods per plant, number of seeds per pod, weight of 100 grains, final population of plants, number of days of the emergency to flowering, number of days of the emergency to harvest, height of first pod insertion, height of the final pod insertion and degree of the down. Analysis of variance was performed, the repeatability coefficient was estimated and the number of repetitions was calculated. Experiments with five replicates identify bean cultivars in relation to groups of characters: weight of 100 grains, number of days of the emergency to flowering and number of days of the emergency to harvest; final population of plants, height of the final pod insertion and degree of the down; and number of pods per plant, number of seeds per pod and height of first pod insertion, with respectively 95, 85 and 80% of real accuracy.
Phaseolus vulgaris L.; repeatability; selection; experimental design
