It was in 1924 that in this Journal we noticed that the apparent velocity of the seismic waves is difierent in different directions of propagation. At that time, owing to the lack of suitable materials, the investigation was confined to earthquakes which occured in the Kwantô plain and the middle part of Honsyû. Since 1923 i.e. the year of the great Kwantô earthquake, the scismic observations in our country made a decided progress and now a dense net of the seismic observatories enables us to extend our study of abnormal propagation of seismic wave to the earthquakes occurring all over the country. The present note is a summary of the results of our researches. (1) Method of investigation. The isochronous lines of an earthquake are drawn on a map so as to fit the localities having equal time of arrival of the first impulse of the longitudinal wave. 106 quakes of tolerably large intensity have been investigated for the purpose. The general form of those isobhronals is not concentric circles, but somewhat resembles ellipses or ovals. The direction of the major axis of such an isochronal ellipse or oval is that to which the seismic wave propagates with a velocity apparently greater than in the other directions. Fig. 1 to 12 show the isochronous lines of earthquakes which occurred at several districts of our country. The hodographs of those earthquakes are drawn with times of occurrences and epicentral distances. On these hodographs are measured epicentral distances corresponding to every 10 seconds of the time of occurrence. With those epicentral distances as the radii, concentric circles are drawn with the epicentres as their respective centres. Then, those concentric circles may be regarded as the mean isochronals free from the abnormality of the propagation of seismic wave due to locality or to direction. The deviation of the observed isochronals from the mean, determines the amount and direction of abnormal propagation of the seismic wave. In Fig. 13 is shown a specimen of such mean isochronals (full line) and the observed isochronals (dotted line) of the destructive earthquake which occurred at about 11h 20m (J. S. T.) on 21st., Sept., 1931 in Saitama province. (2) The direction of abnormal propagation. Almost all the earthquakes which occurred in our country show the abnormality in their propagation. The direction in which the seismic longitudinal wave propagates with a velocity apparently greater than in other directions is determined by measuring the deviation of observed isochronals from the mean. We call, for simplicity's sake, such a direction as that of abnormal propagation. The directions of abnormal propagation have been determined from the isochronals of 106 earthquakes which occurred in 1927 and 1931. They are shown on a map by arrows from their epicentres. Fig. 14 and 15 show the distribution of the direction of abnormal propagation of seismic wave thus determined. From the direction of abnormal propagation shown in Figs. 14 and 15 we see that they are perpendicular to the arc of the Japanese islands in Kwantô and the middle part of Honsyû district. They show West and East direction in the western part of Japan, and the direction of North and South in Tôhoku and Hokkaidô districts and on the Pacific coast of the eastern Kwantô district. (3) The geologicol nature and the direction of abnormal propagation. It is a well known fact, that the geological structure of our country is very complex so the velocity of seismic wave differs according to the geological nature of the medium in which the wave propagates. On this point, we may calla ttention to an excellent investigation of Prof. Kusakabe on the elastic properties of rocks.
The author of this paper proposed a theory to explain the mechanism in accumulating such a large amount of charges on the thunder-clouds. The mechanism here conceived quite similar to the main action of an influence machine in that two conductors come in contact in an electric field and then become separated, resulting in the separation of charges on the conductors. In the case of an influence machine, the conductors are made up of the metal sheets on a rototing circular plate of glass and a third metal conductor, generally wire, made the two sheets in contact when they came in a preferable electric field. In the proposed theory, two water particles take the part of the metal sheets and the third metal wire is omitted but its part is played by the direct collisions of the water-particles. At the outset of the paper, the auther discussed on the point that the water particles which constitute a cloud should be equally charged and their radii are equal if they are not very far from each other. In the next of the paper he ciscussed on a special thunder-cloud. He assumed that the water particles are uniform in size and that all the variables are the functions of the time and the distance from a point. These simplifications enabled him to calculate the distribution of the charges in a cloud. In the case of a thunder-cloud having a spherical form, the charge on the cloud surface does not disturb the process of driving electric charge to the surface of the cloud, at which they accumulate. He insists that the theory stands on the assumption that the clouds have a special conductivity due to the collisions of the water particles.
In my previous papers I have statistically studied the individual cases of cyclonic rainfalls in Japan. As the result of my investigation, I have found that the cyclonic rainfalls are usually much affected by the mountain ranges. Dr. T. Okada has proposed in his Text book on Meteorology that the actual regional distribution of rainfall in our island Empire during the visitation of cyclone results from a combined effect of cyclonic and orographic rainfalls. We have affirmed this fact by means of many typical examples which were classified according to the tracks of cyclones. We have found that the orographic effects are of two kinds; the first one is due to the forced ascending motion of air along the mountain side, and the second one is that which results from the horizontally deflecting motion of lower winds caused by the mountain ranges. In the case of the orographic effect of the first kind, heavy rainfalls usually occur in the mountainous region in general while in the case of the second kind heavy falls occur in the plane or the valley near the mountain. The colder surface wind is usually mild and blows in an opposite direction to or at about right angles to the upper wind. Now we may propose that the second kind is a result of the formation of the surface of discontinuity in the atmospheric strata due to the mountain ranges. In the present paper we will consider the heavy rainfalls which lasted over one or several days associated with tropical cyclones, which travelled over Formosa and its neighbourhood. The typhoons, which pass over this district may be generally divided into four classes according to their tracks, that is, (i) typhoon which moves west or north-westward passing along the northern coast of Formosa; (ii) typhoon which moves in the same direction passing through the southern part of Formosa; (iii) typhoon which passes through the middle part of the island from southeast to northwest; (iv) typhoon which travels along the eastern coast of Formosa from south to north. During the visitation of the typhoon of the first category, heavy rainfall often occurs on the western side of the mountain range. A fall of 1034mm, the record of the daily maximum amount of rainfall in Japan occurred at Hunkiko on Augut 31, 1911, associated with a large typhoon of the same class. On the eastern side of the mountain range the precipitation is very slight, owing to a descending motion of ærial current over the mountain barrier. In the case of the typhoon of the second class, the heavy rainfall occurs in the eastern and southern parts, and is rare in the northwestern region. In the case of the typhoon of the third class, the heavy fall occurs in the mountainous region of the southern, northern and northeast parts. In the case of the typhoon of the fourth class heavy fall occurs in the eastern parts of the island. On the occasion of the cyclonic centre situating far from Formosa, the heavy local rainfall associated with southwest wind in summer occurs sometimes in the southern part but in winter the heavy rainfalls at the north extremity of this island.