The objective of this paper was to determine a method to allow the determination of good experimental unit size, using the point of maximum curvature estimated, by algebraic calculus. There is an empiric relation between the coefficient of variation of experimental unit and its size. One example is presented and discussed to make possible the following conclusions: the method offers an algebraic solucion for the point of maximum curvature; the method does not permit different interpretations when the scale of the graph is dispossed differently; the method may present a size of experimental unit for high coefficient of variation, and in this case it is not recommended.
maximum curvature; algebraic calculus; scale of graphic