()4 



slioit-period waves of small ainplitiKk', whioli arc followed hv the in'inripal 

 movements. Often the early prineipal movements earry, superposed on tliem, 

 .-mall short-period waves. Tluse ijenerally disa])pear before the end, and the 

 eartlupiake as it dies ( ut consists of long-j)eriod waves alone. The first tremor 

 and the sul)se(iiient large movements sometimes occur in a manner strongly sug- 

 gestive of the idea that the first, orshort-jieriod movements, are normal waves and 

 the second, or principal nioNeincnts, arc transvers(^ waves (S !2). 



(S). A ])rogre.ssive, tliougli irregidar, lengthening of period can sometimes 

 1)C detected towards the close of tlie disturbance. 



(9). The jieriod of the principal movements is usually from half a second to 

 a second ; but the slioit-jteriod waves which occur at the beginning mav have a 

 frequency of ö to the .second, or more. 



(10). The v>n-ticd n;otion is generally nuirii Ics-; thm the Imri/untal. 



!; (!'_'. < 'üiiipicxity of f lie niotimi. 

 Tile long dujalion and great ciim])lexity of tlu' movements which the sur- 

 face of the ground ]K'rforms during an ordinary carth(piake in Tokio arc no 

 diiubt largely due to the (jiresumablv) considerable distance of the observing 

 station iVom the origin of disturbance. The theoretical coiisidcratifms which 

 have been adduced in ('ha]>ter I lead us to expect that(cxce])t as regards amplitude 

 of motion) earthquakes grow as they travel through heterogeneous media ; and 

 t)l>servations of the .same shock at different places show that this docs happen. 



In a pa])er descriptive of the earth<[uakc of ^larch 8lh 1881,'" which 

 suj)])lied the earliest clear evidence of change in the direction of movement 

 iluring a disturbance, tin' present writer has sMg<;'ested \-arious explanations of 

 tins change. These are: (l)the presence ol' normal and transverse waves toge- 

 ther. Kven transverse waves alone may give non-rectilinear movements (SI), 

 whose horizontal coinjioncnts may also be cur^•cd ])rovided the ]ilane of the wave 

 is not vertical. ("_') .\ pos>ii)ly wide and not very distant origin, giving waves 

 with various a/innniis. (li) The reflections, lefractions, and diffractions which 

 occur along the route. (4j The (lossible simultaneous, or nearly simullani ous, 

 ■ occurrence of two or more .separate eaith'juakcs. This last explanation is not so 

 extravagant as it may at first sight apjiear to be, if we consider that in a distri't 

 liable (as the Plain of Yedo is) to frcqtient cartlniuakes, there must often exist 

 states of critical ci|uilibrimn, reaily to act as centres of eaithi|nake disturbance. 

 The \ibratious of a siiock occurring at one ],oint might therefore, on icaching 

 one of those places, start another convulsion, which would add its effects to those 

 of the first on neighhouring parts of the soil. 



It appears, however, that changes in the direction of movement during an 

 earthquake are usual rather tiian cxcc])tional ; and tluy must therefore be 

 ascribed, in gineral, to the first three of the causes named above — probably to 

 the finst and third in the gieat majority of cases. 



* Trans, of the Seismological Socictj' of Japan, Vol. Ill, ii. I'Jl. 



