No abstract

At present corn is graded organoleptically into “normal” and various “off‐odor” categories. The possibility of classifying corn odor on the basis of gas chromatographic patterns of headspace volatiles was studied using 60 samples of corn and statistical step‐wise discriminant and regression analyses. Three options for subdividing the patterns into variables and several transformations for quantitation of peak areas were explored. To obtain optimum odor classification, five to seven gas chromatographic features were necessary. When applied to additional samples, classification at an error probability P < 0.05 (above 95% confidence level) was obtained for some statistical models. Thus, an objective system is feasible for classification of grain odor.