■KUBI 



84 



and tlie horizontal velocity of transit of the wave. Four .sufficiently distant 

 stations furnish data for determininor the ejjiccnfrum, or surface-jioint vertically 

 above the origin, the vel(x;ity being assumed coustant and the wave-front 

 spherical. A fifth station, on the same assiunptions, if properlv chosen, gives 

 the additional information needful to determine the dejith of the origin. Know- 

 ing the in.stants of arrival at five stations, we have five simultaneous equations for 

 the thi-ee coordinates of the origin, the velocity, and the instant of disturbance at 

 the origin. 



But, in fact, even the fii-st of these .^^chemes, and still mure the second and the 

 third, is generally incapable of sufficiently exact ai)plication to an cart!ii|uake as 

 a whole, on account of the ill defined beginning and long duration of the dis- 

 turbance. By the time it i-eaches a station distant from the origin, the shock is 

 very far from consisting of a single impulse whose time of arrival may be 

 recorded with precision. A glance at any of tlie diagrams described in Chajjter 

 V will show how indefinite the time of arrival must be, whether we determine 

 it by the sensations of an oljserver or by using a mechanical or electrical seismo- 

 scope. The beginnings of motion are usually so gradual that the time shown 

 depends largely on the sensibility of the appliance used to detect disturbance; 

 and in attempting to measure, by time-takers, the interval between the times of 

 arrival of an eartlujuake, as a whole, at two or more stations, we are liable to 

 errors of perhaps ten seconds or more — errors which may even exceed the quan- 

 tities to be mea.sured. It is true that by increasing the distance between the 

 stations we make any given ernir relatively less important ; liut, on the other 

 hand, the character of the complex system of waves will then be more likely to 

 differ at clifferent stations, and the errors of time measurement will thcrel)y be 

 increased. It L« only in violent earthquakes, where stations may be taken at 

 such distances from each other as to give intervals greatly exceeding the dura- 

 tion of the shaking at any one point, that trustworthy results can be obtained. 



The same objection applies with equal force to the measurement of time- 

 intervals by telegraphing automatically the time of arrival at .several .stations, 

 and recording the signals on a chronograph.* 



In a jiajier read liefore the Seismologial Society in February lS<Sl,t the 

 present writer, after pointing out the impracticability of obtaining precise time- 

 intervals by reference to an earthquake as a whole, gave a scheme by which the 

 velocity and direction of transit might be determined provided we could identify, 

 at three or more stations, any one motion out of the group coastituting an earth- 

 quake. It was proposed to put up three (or four) pairs of continuously recording 

 horizontal jiendulum seismographs at three (or four) stations connected electrically 



* Prof. W. S. Ch.iplin had designed and, in fact, prepared an apparatus to carry out this 

 method, just lefore the writer obtained his earliest records of earthquake motions on a moving 

 plate. The character which those records showed the motion to possess led Prof. Chaplin at 

 once to aljaudon his project. 



t Transactions, Vol. Ill, p. 111. 



