DisplayTitle

Sales agreements at first point of sale and downstream in the market chain often specify a minimum value for falling number (FN), typically 300, below which the consignment is discounted. Although extensively used throughout the grain trade, FN is sometimes criticized because of perceived imprecision. For two successive years, FN precision has been extensively studied at USDA Beltsville. In the first year, repeatability precision, as performed according to AACCI Method 56-81.03 and slight modifications thereof (control + 3 treatments), was evaluated under optimal laboratory conditions and operator practices. Best linear unbiased predictors (BLUPs) obtained from the mixed effects ANOVA yielded estimates of FN averages for each treatment and sample. Coefficients of variation (CV) were calculated using these BLUPs and variance estimates. The 2nd year was devoted to ascertaining the inherent variability (sampling error) of the grain within the truck bed using U.S. official inspection probe-sampling protocol. Results for the first year indicated that CVs between 1% and 4% were achieved for all treatments and samples. Small increases to the sample mass and water volume improved precision, whereas the addition of a wetting agent had a negligible effect. In the second year, with test samples consisting of individual probings of truck beds (7 probings per truck) in Washington (4 sites, 10 trucks/site) and Ohio (1 site, 11 trucks), results for Wash. indicate surprisingly low sampling error, with 3 distinct variance patterns depending on site, these being FN variances of 24 (15-46 95% confidence interval for one site), 10 (5-29 CI for a second site), and 0.06 (with no CI estimable due to closeness to zero for two sites pooled). Thus, results of the precision and sampling studies indicate reasonable performance of the FN procedure, from sampling through lab operations.