millimetre in C a tilt of about 2" would be required. Between February 18 and 25 the same deflection was obtained by a tilt of about 0'-27. Because D recorded more motion than C, it is important to notice that this occurred notwithstanding the fact that on all occasions, excepting between February 4 and February 11, C was very much more sensitive than D.

The temperature variations in the cave during 24 hours never exceeded 1.5 C.

(c) The Diagrams.

At first sight, with the exception of places where earthquakes have been recorded, the photographic traces appear to be a series of long straight lines, and the fact that movements have taken place to the right and left of a normal position can in most cases only be seen by looking along their length (see fig. 2, Plate II.). The curves which are thus seen are too long and flat to admit of accurate measurement, and although the films had only moved at a rate of about 6 mm. per hour there is great uncertainty in determining points of inflection. For these reasons both the angular deflections and the periods occupied in describing them are only rough approximations. In the diagrams, Plate III., the observations extending over nine weeks are plotted as a series of curves. The vertical lines indicate noon and midnight of successive days, which are marked with their dates. The horizontal lines, which are 10 mm. apart, indicate seconds of arc. If the curve for C goes downwards from its starting point the movement is equivalent to a rising on the S.E. side, while if the D curve descends this means that the S.W. rises. The angular values for the various deflections are marked on the diagrams, while earthquakes recorded by seismographs but not by the pendulums are indicated by dots. Earthquakes recorded by the pendulums but not by seismographs are shown by short straight lines.

(d) The Movements of the Pendulums.

From the diagrams it is clearly seen that during any week the pendulums have once or twice wandered away from and then returned to their starting point. Because the movements or periods of comparative rest of the two instruments approximately coincide in time, as, for example, during the fifth week, I take it that the cause of these movements is something more general than a warping of the supporting columns. The movement of D or that which is parallel to the dip has usually been greater than that indicated by C or parallel to the strike. For example, it may be

Earthquakes recorded on the photographic trace.....

Local earthquakes not recorded on the photographic trace

Illustrating the Fourteenth Report of the Committee on the Earthquake and Volcanic Phenomena of Japan.

92

imagined that between February 28 and March 3 the dip of the rocks increased and then decreased through an angle of 4′′·08. About the same time the movement at right angles to this was 288. With the exception of a wave indicated by C between February 6 and 7, which had a period of 24 hours, all the other movements have had periods of from 48 to 70 hours. In this respect the movements are strikingly different from those recorded in Tokio, where diurnal waves are very frequent. Another remarkable feature in the records is the entire absence of tremors, which in Tokio on the alluvium often result in producing a photographic trace from 5 to 10 millimetres, and sometimes even more than this, in breadth (Plate II., figs. 5 and 6). Although it is premature to offer an explanation for the Kamakura movements the following statement may be made. As the changes in temperature in the cave were usually too small to be measurable it is not likely that the wandering of the pendulums can be attributed to such a cause. Any effect that the heat of the sun may have had upon the face of the cliff in which the cave is situated, in raising its temperature or by withdrawing moisture from its surface, would probably be diurnal in its character. The only sunshine records which I have taken commence on February 25. From that day to February 28 there were 28 hours of sunshine which was followed by dull weather until March 11. Although the end of the period of sunshine was followed by a great movement, considerable movements occurred during the comparatively cloudy weather. Because the records are few this observation, however, carries but little weight. Rain, which only fell between February 8 and 9, and again between March 5 and 6, does not show any connection with the movements. Notwithstanding these observations, since the wandering of pendulums in Tokio, as will be shown later, is apparently connected with the movement of subterranean water, which in turn is related to percolation from the surface, it does not seem unlikely that the movements at Kamakura may also find a partial explanation in a somewhat similar cause. The only other explanation, which, however, has not yet been verified, is that they result from rock crumpling which is still in progress, and for this reason the greatest motion is parallel to the dip.

Conclusions of practical importance which I arrive at are that although pendulums which will record a tilting of 03 are sufficiently sensitive to be used on the Tokio plain, an instrument of much greater sensibility is required to study movements on the rock, and, further, that all who have to carry out physical investigations requiring a steady platform will gain great advantage by installations on the rock where tilting is small and diurnal movements and tremors are not appreciable to instruments such as I have employed.

(e) Earthquakes.

The earthquakes which have been recorded by the Kamakura instruments in 1893 are as follows:

The following 26 disturbances are clearly shown upon the photographic traces. In addition to these there are a number of slight irregularities with amplitudes of 1 mm., or under, which have been omitted, first, because they are very small, and secondly, because they might or might not be due to earth disturbances. While the observations were going on, that is, between January 14 and March 18, by means of seismographs in Tokio, 21 shocks were recorded. These were disturbances that were felt in Tokio, and it is known that several of them were also felt in Kamakura. It is probable

that had a seismograph been placed in Kamakura all these shocks and possibly a few others would have been recorded. The horizontal pendulums, however, have only recorded three of the Tokio series, and to the remaining 18 they have been insensible. On the other hand, they have recorded 23 disturbances which the Tokio seismographs have failed to record. Although I have not yet had time to fully analyse the records of earthquakes given by the Tokio pendulums, from what I have seen of them I know that the results will be similar. On many occasions I have watched a horizontal pendulum while a sharp disturbance lasting from 15 to 30 seconds has been taking place. All that happens is that there is a slight elastic switching in a vertical direction of the pointer at the end of the boom. The boom does not swing, and I have not observed a blurr in the photographic trace. (Plate II. fig. 2.) On the other hand, whenever an earthquake instead of simply producing elastic vibrations throws the surface of the ground into undulations, the pendulums behave most erratically. They do not swing, but they are forced first to one position and then to another. Now and then they may pause for two or three seconds, but only to start, perhaps, more vigorously than before. It is interesting to watch a seismograph writing an earthquake, but a horizontal pendulum actuated by earth waves is one of the most attractive sights that a seismologist can witness. When a seismograph is disturbed you feel the motion that causes it to move, and in two or three minutes all is ended; but when a horizontal pendulum is disturbed nothing is felt, and its spasmodic movements may continue for one or two hours. Fig. 3, p. 109, shows what is almost continuous motion for 5 hours 24 minutes. Already I have had the good fortune to see this phenomenon five times. On the last occasion, March 22, at 7.27 P.M., I sent a messenger to call my colleague, Mr. C. D. West, who arrived some 15 minutes later, when we watched the big boom stopping and starting from various positions for 1