2003 年 11 巻 1 号 p. 103-111
Soy sauce and tamari have traditionally had different quality features in different localities. Attempts at improving quality through multiple regression analysis of the relationships between quality and the complex six-month-long fermentation process have not proved to be entirely appropriate. MTS was applied to 100-kL samples of the refined and unrefined product, derived from 20 to 40 tons of raw material. The process was monitored to analyze the relations of process parameters to koji performance (strength value, pH, moisture, bacteria count), the nitrogen utilization factor in the unrefined solution, and the aroma of the final product. From the values of these characteristics, they were divided into a unit group and a signal group. Valid characteristics were selected by the inverse matrix zero-point ratio method, and level trends and signal values were estimated. The results were consistent with known fermentation properties. MTS proved to be a highly effective method of evaluating the process on the basis of various combinations of fermentation conditions.