The objective of this research was to study the regularity of sorghum plants within row, its relationship to grain yield and its effects on statistical analysis. The sorghum variety used was 'BR 304' and the experiment was conducted at the Federal University of Santa Maria at two seeding dates (12/07/05 and 01/05/06) using six plots, three plots at the seeding one date and three plots at the seeding two date. Therefore, the total of plants evaluated was 864 (36 plants/line * 4 lines/plot * 3 plot/seeding date * 2 seeding date). Each plot consisted of four rows (spaced 0.5m) totaling 36 plants. For each plot, the grain yields of each individual plants was measured (Y) and the respective row space occupied by that plant (X) identifying the row (1 to 4) and the number of the plant in the row (1 to 36). For each variable (X and Y) the following tests were conducted: randomness around the mean, Lilliefors normality and homogeneity of variances, for each row and plot as well as the normality for each established plant spacing class (bellow the standard, standard and above standard). Failure on the propositions (normality, randomness and homogeneity of variances) of the variable spacing among plants did not affect the propositions for grain yield of these plants. A covariance analysis for sorghum grain yield, as function of spacing among plants , should be used when spacing between plants is big or when plants are irregularly spaced within row.
Sorghum bicolor L.; normality; aleatority; homogeneity
