In order to know the cause of difference in the expectation of life (ex) by prefecture, relationships between ex (e0, e20, e40, e65) by prefecture in every five years from 1965 to 1980 and constituent of river water (19 items) were examined by simple correlation, partial correlation, multiple correlation and multiple regression analysis, and the following results were obtained. 1) Correlation coefficients between ex and pH, Ca, Mg, CO2, Ca/SO4, Ca-Na were positive and significant, and those between eX and SiO2, SO4, SO4/CO3, Na/CO were negative and significant in one or some age of year. These correlations, excepting PH, were relatively strong in female and old age and showed tendencies of getting weaker year by year. 2) Of the partial correlation coefficients excluded influences of the items whose simple correlation coefficients were significant, those between ex and SO4/CO3 were chiefly significant among others until 1975, so were those between ex and PH in 1980. 3) In the multiple regression analyses of ex on items by forward selection proce dure, SO4/CO3 were selected for the first variables in the most cases until 1975, so were pH in 1980. Above results suggest that the relation of SO4/CO3 to the difference in ex by prefecture was strong, especially for female and old age, getting weaker every year, and making stronger that of pH recently.
Using the male data in 1965, the authors investigated the validity of life lost as a health indicator in eighteen countries and each prefecture of Japan. Moreover, we deviced a new health indicator "Life Gained" meaning age-corrected life gained and investigated the validity of the new indicator as a health indicator. The procedure for calculating the crude life gained is as follow : Crude Life Gained=Σ The results obtained were as follows : 1) The relationships between life lost and life span, infant death rate and corrected PMI were markedly high, but the relationships of life lost to corrected death rate and average death age were not so high. 2) The difference of corrected life lost and crude life lost was not so remarked. 3) The age of 80 years or so was adequate for the target years of age for life lost. 4) The new health indicator "Life Gained" proposed here, was very valid for the estimation of health level in the areas.