A corn density calibration was developed for a near-infrared transmitt
ance instrument (Infratec 1225). The calibration sample set included 9
6 corn samples grown in various locations of the United States. Sample
s were selected by principal component analysis (PCA) from a larger se
t of 410 samples, representing the 1986 through 1992 crop years. Sampl
es for instrument and temperature stabilization were included in the c
alibration. The partial least squares calibration was then validated w
ith 35 randomly selected samples not in the calibration sample set. Wi
th 14 PCA factors, the standard error of calibration was 0.0173 g/cm(3
), and the standard error of prediction was 0.0164 g/cm(3). Fourteen f
actors were required because the first 12 reflected the physical corre
lations of density to protein and density to starch. This calibration
gives the Infratec analyzer capability to measure moisture, protein, o
il, starch, and density simultaneously on an unground sample.