Elastic Waves, with Seismological Applications. 79 



horizontal motion ; but, as I believe, it is the vertical motion 

 ■we must look to specially. 



Another point brought out strongly by Sekiya's analysis 

 is that in no case has he found the vertical motion pre- 

 cede the horizontal motion. The vertical seems always to 

 show itself later. It is certain, however, that earthquake 

 sounds are often heard before the earthquake shock is felt. 

 This simply means that the big earth shiftings which affect 

 our seismographs are preceded by rapid vibratory motions 

 which, however large they may be, cannot have any mechan- 

 ical effect on the instruments. The case is exactly similar 

 to what happens if we pass alternating electric currents 

 through the coil of an ordinary galvanometer. No matter 

 how sensitive the galvanometer, or how intense the alternating 

 current, — so long as the alternation is rapid enough, no effect 

 is observed on the galvanometer needle. So it cannot fail to 

 be with ordinary seismographs as regards rapid vibrations. 

 I doubt if a seismograph, mechanically capable of registering- 

 vibrations occurring at even so slow a rate as 10 per second, 

 has been as yet imagined. It is very questionable also if 

 those sinuous records which the seismograph tracings show 

 as precursors of the large slow waves really indicate what is 

 taking place in the soil. For, exactly as very rapid vibra- 

 tions will not show at all on the seismograph trace, so some- 

 what less rapid vibrations will not show to their full. 

 There must always be a lagging of the record behind the 

 motion recorded. Thus before a given motion has its full 

 effect on the seismograph, the rapid reverse motion may set 

 in and prevent anything like a complete record. Not until 

 a comparatively slow " swing-swang " of the ground takes 

 place can we hope to have a tracing even approximately true 

 as to amplitude. It is therefore well, I think, that seismo- 

 logists should bear this point in mind. It is highly probable 

 that an earthquake is preceded by rapid vibratory motions. 

 That we should expect ; and the early sinuosities of earthquake 

 tracings certainly suggest the same. But that these sinu- 

 osities can be taken as an approximate representation of the 

 amplitudes or periods of the rapid vibration to which they 

 are due may well be matter of grave doubt. 



In conclusion, I would draw special attention to the fol- 

 lowing point which seems to be of some importance. 



In the discussion of the propagation of seismic disturbances 

 through the earth's crust, a clear distinction should be drawn 

 between purely elastic and QUAtii-elastic phenomena. So lono- 

 as the materials constituting the earth's crust are not strained 

 distinctly beyond the limits of elasticity, we have to do with 



