Takata is one of the localities noted for heavy snow fall. Using the 15 years, observations beginning with 1909, the author investigated the relation between the depth of the deepest snow at Takata and the monthly mean of the atmospheric pressure for each winter months from Oct. to March for the continental stations, Changchun in Manchuria, Tientsin near Peking, Hankow in Yang-tze Valley in the same winter. The correlation coefficients between both variates have been calculated and found to be small, indicating the relation between them is not conspicuous for the months Oct. to Dec. and Feb. to March. For the month January, however, the correlation coefficients are strikingly large as shown in the following table. These results show that, the depth of the deepest snow at Takata is closely related, with the monthly mean pressure for January over the continent being closest with that over northern part i.e. Manchuria and rather less with that over southern part i.e. Yang-tze Valley. This conclusion is apparently contradicted by the fact that, when a cyclone passed over the southern part of Japan Sea near the coast of Hokuroku, and consequently the pressure distribution becomes higher towards SW and lower towards NE, then at Takata and in its vicinity almost always a heavy snow-fall takes placc. But by the author's opinion, this is not true contradiction to the result obtained above, because in the latter case the monthly mean of the atmospheric pressure is treated and in the former case one concerns with a single current phenomenon.
The present author gives results of statistics on the duration of rainfall observed at the Nagano Meteorological Station. Results are shown in 4 tables. (p.) The first table contains frequencies of rainfall which set in A. M. (and which set) in P. M. The second table contains the result of summation of rains beginning between Oh-6h, between 6h-12h, between 12h-18h and between 18h-24h. In these tables, monthly values for April to October are shown. These data are taken from the records of a selfregistering raingauge. In winter pluviograph is not generally in good working order, so that results for November to March are not given here. In these tables the small figures in bracket are numbers of thunderstorms in corresponding monthes. Small figures at right foot are numbers of rainfall, amount of which is less than 0.1mm. The third table gives the frequency of the period of breaking of rainfall. The author observed the time between the beginning of a rain and its next breaking epoch, or between two epochs of breaking, and numbers of events corresponding to 0-5 min., 5-10 min., …… are given. To get rid of the ambiguity of determination of such breakiug epoch, the author compares these results with the smoothed values obtained by a method of mean 1/5(a/2+b+2c+d+e/2). For example, in the second column for April corresponding to 25, 30, 35, 40 and 45 min. (in the first column), numbers 1, 8, 4, 15 and 7 are given. From these numbers a value 1/5(1/2×1+8+2×4+15+1/2×7)=7 is shown as a smoothed value in the third column to correspond to 35min. Fat figures in this table are maximum values which are greater than cousecutive ones before and after. In the forth table, data about heavy raius and relation between its intensity i and duration t are shown. Most of product i√t=n (Köppens, Das Wetter P. 169) lies between 2 and 3 which is very near to Wussow's1)value √5. But in case of shorter duration the products are much greater than this value √5, this case resembles Fassig's2) result observed in America3).