In none of the civilized countries of the world we find such a high morbidity rate of typhoid fever as in Japan. I think, therefore, that it is a very important problem in the field of public health of our country to elucidate statistically the relationship between the morbidity rate of typhoid fever and factors of vital statistcs affecting it according to geographical districts. For this purpose the method of partial correlation is adopted. The factors taken into consideration, and their notations, are given below: t=Morbidity rate of typhoid fever Number of cases of tvnhoid fever/ Population. - 10, 000 d=Densiry of population=log (Population/ Area of land in Km2) s=Sex ratio of population=Number of male to 100 female a=Mean temperature (chiefly at the site of a prefectural office; if not, in the neighbourhood of it.) b=Distribution of wéalth Amount of postal savings / Population c=Urbanization=log. (Population in cities/ Total population. -1, 000) g=Plevalece of education Number of graduates of girls' high and traing schools/ Female population f=Number of people per family= Ponulation/ Number of families h=Age composition of population = Population of 15-44 yeafs of age / Total population p=population The data utilized are those of the year 1930, when the national census was taken. 1. Density of population (d) and morbidjty rate of typhoid fever (t). It is not unreasonable to expect an existence of a positive correlation between the morbidity rate and the density of population. But a negative, insignificant correlation between d and t was found, thus; If the Influence of other factors is eliminated, the above coefficient turns out to be a positive, but still being insignificant, This shows that the density of population itself is not responsible for the increase of the morbidity rate of typho: d fever. The original negative correlation of the first order between t and d must be interpreted as due to the other factors. Among such factors mean temperature (a) is important. The denser the population, the higher the mean temperature, and the higher the mean temperature, the lower the morbidity rate, Thus we see that the coefficient of the first orber comes out negative. 2. Numbet of people per family (f) and morbidity rate of trfihnid fever (t). If the correlation between t and f are computed, the following coefficient is obtained, which shows that prefectures havingr a larger number of people per fain-ily have also a higher morbidity rate. But if the other factors are held constant, the coefficient of partial correlation becomes insignificant, This result indicates that the numder of people per family is not, an important factor influencing the morbidity rate of typhoid fever. Hence the original correlation of the first order must be due to the other factors. Among these factors the mean temperature is of importance. which shows that prefectures having a greater number of people per family should have a lower mean temperature. And the lower the mean temperature, the higher the morbtdity rate, as we saw before. 3. Urbanization (e) and the morbidity rate of typhoid feaer (t). In Japan, urban districts have generally a higher morbidity rate of typhoid fever than rural areas, as follows: which is not significant, but still positive. But if the influence of other factors is eliminv, ted, we obtain the coefficient, Obviously this coefficient depicts a presence of significant correlation between these two variables t and c. That is to say, urbanization itself exerts an important influence upon morbidity rate, and when we think of the incomplete sewerage in the majority of cities of Japan at present, it seems reasonable to expect that the completion of perfect systems would greatly contribute to the reduction of typhoid fever cases, as in the number of examples in European and American cities.