STRUCTURE OF THE EARTH HODGSON 285 



second is propagated by a transverse motion of adjacent particles 

 somewhat after the manner in which a water wave travels ; the floating 

 debris shows that the water particles move at right angles to the 

 direction of propagation of the wave. 



The difference in velocity depends on the elastic properties of the 

 body through which the waves travel. For short distances it is a 

 sort of neck-and-neck race. No laboratory is large enough to enable 

 the waves to separate sufficiently for the fact to be established experi- 

 mentally. So for many years there was no experimental verification 

 of the two types of waves. After earthquakes were registered at con- 

 siderable distances, it was found that a second burst of energy showed 

 on the record among the dying vibrations of the first impulse (see 

 fig. 2). The greater the distance from earthquake to station, the 

 greater was found to be the separation of these two impulses. It was 

 finally established that these were the arrivals of the two types of 

 waves which had left the origin of the earthquake together but had 



-nA^Wi^'*f<*>/^>'vv«^' 



vA^iW<AA^^ 



Figure 2. — Section of a selsmogram showing the P and S phases. 



traveled at different speeds. The difference in time of arrival is thus 

 a measure of the distance. 



The two types of waves of which we have been speaking are 

 designated, respectively, P and S waves by seismologists since, in 

 point of time of arrival, they are primary and secondary. It is readily 

 seen that, if observers possess a timetable showing the time required 

 for the P wave to reach various distances on the earth's surface and 

 a similar one for the S wave, a table could be prepared showing the 

 relation between S — P times and the epicentral distances. The first 

 table of this sort was prepared by Oldham in 1900, and since then a 

 steady improvement has been effected. Such compilations are called 

 time-distance tables or, if plotted on graph paper, time-distance 

 graphs. Given tlie table, it is easily seen how the distances from 

 each station to the epicenter can be deduced. Now, if it be known 

 how far the epicenter lies from any station, then the epicenter must 

 lie on a circle drawn about that station with the proper distance 



