Principal components as indicators of body fatness and subcutaneous fat patterning.

Principal component analysis was performed on the skinfold data of contrasting groups of 50 New Guinean men, 47 New Guinean women and 138 Italian men. The common components of fatness and limb-trunk fat patterning were identified and the relationship of component scores to fatness and fat content, determined densitometrically, and to fat distribution as evinced from simple skinfold ratios, were ascertained. The interpretation of the components appeared correct: correlation coefficient between fat mass and first component scores was 0.89, P less than 0.001, and between abdomen/thigh skinfolds and second component scores was -0.78, P less than 0.001. However, log sigma skinfolds was as good an indicator of fatness and not all limb/trunk skinfold ratios were highly correlated with second component scores. The identification of the simplest and most rigorous method of describing fat patterning and distribution and the factors affecting them require further investigation but principal component analysis is proving useful.