June 19, 1S84J 



NATURE 



175 



sentativc of a very large proportion of the earthquakes 

 which occur so frequently in the Plain of Yedo. Earth- 

 quakes of this class do no damage to buildings, but 

 they are strong enough to make their presence felt by 

 the shaking and creaking of houses, and even, in the 

 night, to startle residents out of sleep. Lamps and other 

 pendulous bodies are frequently set into considerable 

 oscillation through the long continuance of the disturb- 

 ance, the period of some consecutive vibrations of the 

 ground being nearly uniform and equal to the free period 

 of the lamp. The shaking lasts rarely less than one and 

 sometimes as much as ten minutes. 



In some cases, however, the amplitude of the earth's 

 motion is considerably greater ; occasionally it rises to 

 5 and even 7 mm. With such an amplitude as this, 

 and with the ordinary frequency which the earthquake 

 waves have, the shock is more or less destructive — walls 

 are cracked and chimneys are overthrown. The writer's 

 observations do not include any earthquake of first-rate 

 violence, but they show by several examples that in the 

 alluvial soil of Tokio a sufficiently alarming and even 

 damaging earthquake may occur, in which the range of 

 horizontal motion is less than a single centimetre. 



In the Yedo earthquakes the vertical motion is generally 

 much less than the horizontal, and, as a rule, forms an 

 unimportant part of the disturbance. 



Fig. 7 is a copy, reduced to about half size, of the record 

 of one of these more considerable earthquakes (on March 8, 

 1SS1), traced by a pair of horizontal pendulums on a re- 

 volving plate. The inner circle shows the N.S. compo- 

 nent, and the outer circle the E.W. component of the 

 displacement. The records begin simultaneously at the 

 points marked a and a respectively, and extend in the 

 direction of the arrow over nearly two complete revolu- 

 tions of the plate. At the point marked c in the outer 

 circle, when the earthquake oscillations were slowly dying 

 away, the writer (who happened to be present) withdrew 

 the plate, to prevent the later portions of the record from 

 confusing the earlier portions. By this time the earth- 

 quake had lasted for two minutes and a half, and some 

 200 vibrations had been registered. The motion, as re- 

 corded, was exaggerated in the ratio of 6 to 1 ; hence 

 in the diagram as it appears here the displacements are 

 nearly three times the natural size. 



For the sake of exhibiting some interesting features of 

 this earthquake more clearly, the records of the two com- 

 ponents during the first twenty seconds of visible motion 

 have been reproduced in the centre space of the diagram 

 in such a manner that simultaneous parts of both are on 

 the same radius. The short radial lines mark seconds of 

 time. It will be seen that for three seconds the motions 

 were very minute ; then the E.W. seismograph became 

 pretty sharply disturbed, but the other component was 

 scarcely visible until the tenth second from the beginning. 



During the tenth and eleventh seconds the phases of 

 the two components agree in the main, but they soon 

 diverge : and in the fifteenth second, when the motion is 

 greater than at any other part of the whole disturbance, 

 they differ by about a quarter of a period. Hence at 

 that time points on the earth's surface were vibrating not 

 in a rectilineal path but in loops. This is strikingly shown 

 by Fig. 8, which shows the path (exaggerated in the ratio 

 of 6 to 1) of a point on the earth's surface, during three 

 seconds at this epoch in the disturbance. Starting from 

 p at 137 seconds from the beginning of the earthquake, a 

 surface particle described the tortuous path shown in the 

 figure, and reached q three seconds later. Similar rapid 

 changes of phase-relation occur throughout the rest of the 

 disturbance, and in the slowly dying oscillations with 

 which the earthquake drew to a close the writer noticed 

 one of the pointers moving vigorously when the other was 

 nearly at rest, and vice versa. 



The evidence, first clearly given in this earthquake, of 

 the non-rectilineal character of the ground's motion, was 



confirmed by very many later observations. In fact in 

 every case where the records were sufficiently large and 

 well-defined to admit of a satisfactory comparison of the 

 phases of the two components, the same thing was exhi- 

 bited. And not only in those cases, but even in very 

 minute earthquakes, instruments having two degrees of 

 horizontal freedom, such as the duplex pendulum, showed 

 in the most direct manner that the earth's movements 

 consisted of a multitude of twists and wriggles of the 

 most fantastic character. 



An excellent example of a still sharper earthquake is 

 given in Fig. 9 — a record (reduced to half size) given by 

 two horizontal pendulums with a multiplying ratio of four 

 to one on a plate which was turning once in fifty-four 

 seconds. The beginning of motion can be detected on the 

 outer circle at a. At b and the corresponding point b' it 

 increases somewhat suddenly, and during the next few 

 seconds we have the principal motions, followed during 

 many minutes by a long trail of lesser irregular oscilla- 

 tions, in which a marked lengthening of period may be 

 detected towards the close. To allow the phase-relation 

 during the principal part of the shock to be examined, 

 lines (numbered 1 to 16) have been drawn by the aid of 

 templates through corresponding points in the two records. 

 An examination will show that the phase-relation changes : 

 in fact when the two components are combined the move- 

 ments are found to be loops, agreeing very closely with 

 the larger loops of Fig. 10, which is a "static" record of the 

 same earthquake given by the duplex pendulum. In a 



part of Fig. 10 the motions are so numerous and so much 

 distributed over all azimuths, that the film of lamp-black 

 has been completely rubbed away from a portion of the 

 plate which received this record. 



It frequently happens in the record of an earthquake 

 that the motions which are first recorded are rapid vibra- 

 tions, of short period and small amplitude, which are 

 immediately followed by larger and less frequent move- 

 ments. Sometimes, indeed, the former appear as a ripple 

 of small waves superposed on larger ones. But in all 

 cases where the short-period waves can be detected they 

 die out early, and the later part of the earthquake consists 

 of relatively long-period waves alone. Records of this 

 class are exceedingly suggestive of the arrival of first a 

 series of normal waves (that is, waves of compression and 

 extension), constituting the rapid tremor, and then a series 

 of transverse waves (that is, waves of distortion), forming 

 the principal motions of the earthquake. 



In fact it is difficult to explain the rapid changes of 

 phase in the two components, or, in other words, the 

 curved character of the horizontal movement, which most 

 if not all the recorded earthquakes exhibit, otherwise than 

 by supposing that the principal movements are transverse 

 waves occurring in a plane not very much inclined to the 

 horizon, and this conclusion is supported by the smallness 

 of the vertical component. 



It is true that the appearances presented by the dia- 

 grams could be accounted for by assuming the presence, 



