on Earthquakes in Japan. 369 



our probable origin are highly faulted, and bear strong evi- 

 dence of recent elevation. The jolting movements felt in an 

 earthquake resemble very much what one would expect to be 

 produced by the tearing open of a fracture or the sliding of 

 two irregular surfaces over each other. 



It has been usual to deduce the angle of emergence from 

 observations of the horizontal and vertical components of the 

 earth's motion. If, however, we accept the possibility of dis- 

 tortional vibrations playing an important part in earthquake 

 phenomena, the accuracy of this mode of estimating that angle 

 is very questionable. It is probable that in many cases the 

 vertical motion may be due simply to a vertical component of 

 transverse movement. When we take this into consideration, 

 it is at once evident that some means must be adopted for the 

 purpose of finding out to which set of vibrations the vertical 

 motion is due. This can be done by writing on a continuously 

 moving plate a record of the whole movement, and observing 

 the vertical movement corresponding to each individual hori- 

 zontal movement. We must not conclude, however, that when 

 the vertical movement coincides with the first indicated 

 movement it is a genuine vertical component of a normal 

 vibration, unless distinct evidence of the presence of trans- 

 verse vibrations is afterwards observed. In order to form a 

 reliable estimate of the nature of the vibration, and there- 

 fore of the angle of emergence, the direction of propagation as 

 well as the direction of vibration must be independently 

 observed at several stations. The direction of propagation 

 can only be reliably obtained from observations of time of 

 arrival. 



An element of uncertainty in time-observation when taken 

 by clock-stopping apparatus is the prolonged duration of the 

 shock, and hence the difficulty of making all instruments act 

 at the same part of the shock. In consequence of this, it 

 becomes necessary either to make the stations at considerable 

 distances apart, or to place them so close together that they 

 can be conveniently put in electric communication, and then to 

 mark particular instants on the different chronographic plates 

 used for receiving the record. In this way, if instruments can 

 be obtained which will not of themselves considerably modify 

 the nature of the record, either the beginning of the shock, or, 

 if that is too gradual, some particular prominent feature of 

 the shock, may be selected on the different records, and from 

 it and the time-signals the interval between the two stations 

 obtained. For direction of propagation three stations are of 

 course necessary; and for the determination of origin two sets 

 of three, if the elements of each set be close together, are 



