Collaborative evaluation of universal calibrations for the measurement of protein and moisture in flour by near infrared reflectance

Publisher Website
Google Scholar
Abstract
Summary: Calibrations which had been previously determined for the analysis of protein and moisture in U.K. and Irish milled straight run white flour were set into nine near infrared (NIR) reflectance instruments of the same model situated in different laboratories. One set of samples was circulated to adjust the bias (intercept) which has a component specific to each instrument then a second set was circulated 2 months later to evaluate the adjusted calibrations. Kjeldahl proteins were determined on the first set by seven of the laboratories and on the second set by one reference laboratory for direct comparison of that method with the NIR method and in order to assess the accuracy of NIR as well as precision.On the second set of samples the mean NIR results for the different laboratories varied from 10.92 to 11.30% compared with a mean for the Kjeldahl of 11.03%. The accuracy of NIR calculated separately for each laboratory as the standard deviation (s.d.) of differences between NIR and Kjeldahl from the reference laboratory was between 0.13‐0.33% while the precision or repeatability (within‐laboratory) was 0.07%. For comparison the accuracy of Kjeldahl, calculated separately for each laboratory in the same way as for NIR, was 0.10% ‐0.27% and the precision (within‐laboratory) was 0.04%.In the case of moisture, the NIR results for the second set of samples had mean values ranging from 13.08 to 13.28% compared with a mean for oven drying of 13.13%. The accuracy of NIR was 0.20‐0.29% with no significant biases and the precision was 0.05%.

This publication has 1 reference indexed in Scilit: