Abstract
Near-infrared spectroscopy (NIRS) is a very useful and powerful method to determine protein content in rice grain. Protein content estimated by NIRS was varied some factors, such as sample cells, grinding conditions, contamination of hulls, etc. But precise evaluation of the influence was not done. In this report, we described protein contents in brown rice measured by NIRS in some conditions in 1993 and 1994. The samples used were gently gifted from many agricultural stations all over Japan. The multiple regression equations established between protein contents by chemical analysis and those of NIRS gave the good multiple correlation coefficients (larger than 0.98). And also, standard errors of calibration and standard errors of prediction were smaller than 0.15% indicating the equations were satisfactory. We tested whether some factors affected or not protein contents estimated by NIRS. Difference of sample cells and measurers, and contamination of hull did not affect protein contents significantly. When powder sample was repeatedly analyzed with one packing (repeatedly used) or with repacking another part of the sample, these handlings did not either. Measurement with and without cleaning of the mill did not. Average powder size ground by different methods of grinding (different screens in Udy mill and a coffee mill) was greatly affected, but it did not affect protein content except the coffee mill. These results indicated that many factors described here did not affect protein contents estimated by NIRS.
