The author tried to analyse the quasi-periodic change of air temperature. The duration and amplitude of the individual fluctuations are nearly determinate, yet subject to some small variation with time. The following method can be applied to other natural phenomena. Suppose yn denotes air temperature or other natural phenomena as a function of time expressed by the following expression: where n is an integer, τ a certain time interval, θ time epoch, and T the period which varies with n. Now we define L and S as follows: accordingly, From (1) and (2) we get If T(nτ)=2τ, Therefore by choosing τ properly, we can anapyse yn into two parts, i. e. periodic and nonleriodic terms. Applying the above method, the author has found five systematic terms in the change of the monthly mean temperature over Japan proper and its vicinity, two of which seem to reign over the whole northern hemisphere. Some considerations are made on their physical meaning.
The coefficients of correlation between the yield of Italian millet and the weather factors (i. e. the monthly mean temperature, monthly amount of rainfall, monthly total number of rainy days and monthly total hours of sunshine) have been calculated by the method reported in a previous number of this magazine, for each prefecture in Japan, for each month during the cultivating time. The correlation coefficients calculated are shown in tables 1, 2, 3, and 4 and the following relations are recognized. 1. The correlation coefficients between the yield and the air temperature in the northern part of Japan are larger than any other correlation coefficients, and they show that the warm, sunny and dry weather is favourable for the cultivation of Italian millet. 2. The rainy weather is favourable for the cultivation of Italian millet in the central provinces of Japan, even though tshe hours of sunshine are somewhat few. 3. The dry weather is favourable for the cultivation of Italian millet in the southern part of Japan, provided that it is not so severe as to cause a drought.