Near infrared spectroscopy has been used to determine the chemical composition of food stuffs, i, e., the protein, fat, and moisture content. Recently in the U. S. A, this technique was applied to the estimation of human body composition, and a portable instrument was developed. This study was designed to determine the prediction equation of body fat for Japanese people. 69 men and 52 women, aged 18 to 58 years, covering a wide range of percent fat levels, physiques, physical activity levels,

more »

... d body fat estimated by hydrostatic weighing (HW), skinfold thicknesses(SF) and near infrared spectroscopy. Near infrared interactance spectra were measured on the anterior midline of the biceps halfway (between the anticubital fossa and acromion), the spectral data from which gave best correlation with HW and SF compared the other sites, using the wavelength of 947 nm. The spectral values and percent fat values obtained from HW about 1/2 of subjects were used to develop a prediction equation. This equation was then used to predict % body fat of the other 1/2 subjects. The correlation coefficient between % body fat as predicted by the near infrared method and as predicted by the HW technique was 0.88 (p<0.001, SEE=3.2). This correlation value was similar to the value between % body fat as predicted by HW and as predicted by SF. In the American data, the correlation coefficients were higher when the other variables were added ; height, weight, age, but in the Japanese they did not almost change even when the other variables were added. There was not a difference between the spectra of right and left arm values. In the results, the prediction equation of % body fat for the Japanese population was determined as follows : % body fat =54.14-29.47•~(the spectral data at 947 nm)•kr=0.88, p<0.001, SEE=3.2•l (for right arm).