Fig. 4. Frequency of errors in predicting observed values by sieve size class using fractal, log-normal, and Weibull distributions.