Multivariate analysis as a tool to predict bread volume from mixogram parameters

We have shown that the baking quality expressed as bread volume of 21 different wheat cultivars can be predicted with 91% explained variance. A 35-g mixograph was instrumented and interfaced to a computer for data acquisition. The obtained mixograms were evaluated with a data processing program extracting 12 parameters from each mixogram. The results were correlated to bread volume with multivariate partial least squares analysis. Five mixogram parameters were found optimal. One parameter, buildup, the difference between the mixing torque at optimum dough development and the mixing torque after the water absorption, was found especially important as it alone explained 77.9% of the variation in bread volume. Including the protein content among the five mixogram parameters resulted in a small increase from 90.9 to 92.8% of explained variation in bread volume. The protein content alone explained 55.0% of the variation in bread volume.