A large number of variables were examined simultaneously for 43 normal subjects over the four irrigations of a caloric test. Care was used for every step of the testing procedure and data analysis to eliminate as much of the variance in the caloric responses as possible. The normality of each variable's distribution was examined using the Wilk-Shapiro W test and corrected if necessary by the best of several transformations. The means, standard deviation, and 95% confidence intervals of the resultant data were derived. Statistical tests of temperature, sidedness, and directionality were done on the variables and several important sources of variance were found and explained.