Using calibrations developed for fine milled meat and bone meal on spectra measured on non-milled samples

A database of 316 near infrared (NIR) spectral data from finely milled meat and bone meat IMBM) samples were analysed in a Foss NIRSystems 6500 SY-11 using the rectangular 1/4 cup. Calibration equations were developed for the prediction of crude protein and ash, two of the parameters used for quality control and payment in commercial MBM meat transactions (R-CP(2)= 0.96, SECVCP= 1.31; R-2 Ihll = 0.94, SECVashes = 0.9 1). Ad d itio na Ry, the possi bi lity of u si ng ca li b ratio n s develo ped with f i ne m i Ri n 9 sa m pLes to p red ict spectra obta i ned i n the same instrument and analysed in the non-miLled form, using the rectangular natural cup was examined. Three different standardisation sets were evaluated for optical matching of the samples analysed in the two different modes. The best standardisation matrix was applied to 10 validation samples. The predicted values from standardised spectra of non-miLled samples analysed in the large cups and the predicted values from spectra of milled samples gave standard error of differences (SED) for CP 10.74) and ash (0.62) that were simiLar to, or even Lower than, the standard error of prediction (SEP) values for the same milled samples scanned in the rectangular 1/4 cup (1.28 and 1.11, respectively). The study suggests that NIR equations developed for fine milled products analysed using a small window size may be used in the same instrument for the analysis of non-milled products analysed under a larger optical window.