Forty Japanese college women, the majority of whom were born and live in Kyushu District, were exposed to three thermal conditions in the Climatic Chamber of Kyushu Institute for Design Research in an attempt to examine BERGMANN's rule experimentally. Skin temperatures at six locations (auricle, chest, arm, thigh, leg and toe) were measured with the aid of C-C thermocouple at the ambient temperature of 15°C, 25°C and 40°C with constant relative humidity of 50% and a constant windspeed of 18cm/sec.<br>Larger variances were found in the skin temperatures at the ambient temperature of 25°C, while smaller variances were found in those at 40°C (Table 1). The variances of skin temperatures at the ambient temperature of 15°C were less than or equal to those at 25°C(Table 1). High and statistically significant correlations found among the skin temperatures at the ambient temperature of 25°C contrast strikingly with low and statistically insignificant correlations among those at 15°C(Table 2). The marked difference in the correlationship of the skin temperatures between two thermally different conditions may imply an alteration in the mechanism controlling skin blood flow in the cold environment. At the ambient temperature of 40°C, statistically significant correlations were found exclusively among the skin temperatures in the limbs (Table 3).<br>Twenty somatometric measurements including 4 skinfold thicknesses were taken. Their means and standard deviations are listed in Table 4. To reduce the dimensionality of the somatometric measurements effectively, a principal component analysis was made based on the correlation matrix of 20 somatometric dimensions. The first four principal components having more than average eigenvalue accounted for 72.38% of the total variance (Table 5). The factor loadings for the first four components are given in Table 6.<br>A multivariate regression model was employed to analyze effects of physique in terms of the first four principal components on the skin temperatures at each thermal condition. At 15°C, a null hypothesis that a set of multivariate regression coefficients equals zero matrix was rejected at the 5% level of significance (Table 7), while at the other thermal conditions the hypothesis was not rejected (Tables 9 & 10). Multiple correlation coefficients (Table 8) serve to interpret the result obtained by multivariate regression analysis. At the ambient temperature of 15°C the skin temperature in the thigh is explained about 30% by the first and fourth principal components. Since the first component represents rather "mass" factor than "size" and the partial regression coefficient of the first component is positive (0.411), it can be said that the more voluminous a woman is, the warmer skin she has in the cold environment. Thus, it is concluded that BERGMANN's rule holds for Japanese females in Kyushu District in the cold environment but not in thermally neutral or warm environment.