In the present paper the authors studied statistically on the remarkable thunderstorms which occured in Japan during 11 years from 1924 to 1934. They classified the remarkable thunderstorms into two main types, namely the one being of the heat type and the other of the depression type. The former includes the thunderstorms caused generally by the solar agency. Here we picked up the statistical data of the thunderstorms occured in the four months from May to September, as in this country the thunderstorms of heat type are very frequently experienced in summer season. The thunderstorms which cause the various damages on the electric transmission lines and so on are mainly of this tpye, hence it seems that the statistical investigation of the remarkable thunderstorms of heat type deserves well to the practical purposes (Fig. 1 and 2). The remarkable thunderstorms of the depression type are, on the other hand, very few in number to investigate them statistically. But, roughly speaking, it is concluded that these are frequently experienced at the western part of the Japan proper and generally they pass over the inland sea from west to east covering more than 500 kms of length (Fig. 3).
In this paper the present author studied statistically on the seasonal variations of the days with thunderstorms for almost all of the meteorological stations of this country. In the mountainous part of the Japan proper the thunderstorm of heat type shows an extraordinary predominance. At the coastal part, the summer time does not contribute to the formation of thunderstorms so much as in the inland part, but in the colder season, i.e. autumn and winter, the thunderstorms are experienced very frequently. The Pacific coast of the south of the Japan proper experiences as much thunderstorms in autumn as in summer, while at the Japan sea coast of the northern half of the Japan proper, the seasonal features are shown very markedly. Namely, in the northeastern coast of the Japan sea, ranging from Sapporo (about Lat. 43° N), to Anbetu (about Lat. 50° N), the thunderstorms occur most frequently in September; at the Japan sea coast of the Japan proper, ranging from Sapporo to the south of Akita (about Lat. 39° N), the maximum frequency of the days with thunderstorms occurs in October; the Japan sea coast ranging from the north of Niigata (about Lat. 38° N) to the neighbourings of Wakasa bay (about Lat. 36° N) experiences the thunderstorms most frequently in November. In other words, the Japan sea coast has a marked geographical feature to the occurence of thunderstorm and the month of the maximum frequency travels southwards from Karahuto to the middle part of the Japan proper as the colder season advances. This characteristic feature may throw a light upon the investigation of the mechanism of the formation of thunderstorm on the point of view to show some climatical states preferable to the occurence of thunderstorm.
In the following, it is proposed that Angervo-Petterssen's kinematical analysis of weather charts should be introduced to our weather bureau system. For practical application to Japanese weather forecasting, some remarks are stated. Next Angervo's formulae on the path of a pressure center are proved in different ways. Also Augervo's results on the formation of a depression (or an anticyclone) from a V-shaped depression (or a wedge) are checked in simpler manners.
It is expected (Vernadsky: La Géochimie, p. 50) that fresh underground water dissolves little or no free oxygen in it, as the rain water which is saturated with air loses the dissolved oxygen in soil by the actions of soil materials and organisms. The author could ascertain, on analyzing some water in the Artesian well, that the oxygen content was low (8.9%-1.2% of the saturation). The samples of water were got from several wells in the suburb of Yokohama and in the compound of the Aeronautical Institute in Tokgo.
From the observational data obtained during 1923-1934 at the Aerological Observatory, Tateno, Japan a systematic study of the correlations between the meteorological elements on the ground and the upper winds were made, by a modification of a method due to K. Han'i. Expressing the wind as a vector and taking components in four cardinal directions good results were obtained. In parts IV, V, VI, and VII of the present paper those results which might be useful for long period forecasting are given: as an example the correlation coefficient between the average velocity of the upper wind and the atmospheric pressure was found to be +0.91, and that between the northerly upper wind in summer and the rainfall in November to be about +0.95. In part III the relations hetween the upper wind and the temperature, the monthly variation of the rainfall during one year, and also the solar constant are discussed. Finally the behaviour of the upper winds at Korea and Tateno are compared with.