198 ANNUAL REPORT SMITHSONIAN INSTITUTION, 19 3 4 



these average times are probably accurate to within 1 or 2 seconds, 

 but a point that remains doubtful is to what focal depth they cor- 

 respond. The next step is to compare the average times with the 

 data obtained from detailed studies of the records of individual 

 earthquakes; in this way local variations in the velocities of the 

 waves and differences in the structure of the earth may be revealed. 



WAVES IN THE EARTH'S CRUST 



Some information as to the constitution of the upper layers of 

 the earth's crust in certain continental regions has been obtained 

 from the study of near earthquakes. The records of shocks at dis- 

 tances of less than about 800 kilometers often show not only the 

 normal P and S pulses, but also other pairs of compressional and dis- 

 tortional waves that have traveled in the crustal layers. According 

 to Jeffreys' interpretation of these waves there are at least 3, per- 

 haps 5, layers concerned : the upper layer of granite (which is gen- 

 erally overlaid by 1 or 2 kilometers of sedimentary material) is about 

 10 kilometers thick; the intermediate layers between the granite and 

 the ultrabasic rock (which transmits the normal waves) are together 

 about 20 kilometers thick. 



The existence of the crustal layers not only accounts for the 

 phenomena of near earthquakes, but also goes a long way toward 

 explaining certain observed characteristics of the surface waves 

 which are not in agreement with the theory of waves on a homo- 

 geneous solid. For instance, the simple theory does not account for 

 the long train of waves which characterizes the L phase ; the velocity 

 of Rayleigh waves in a homogeneous earth is independent of the 

 wave length and so the Rayleigh phase should consist of a single 

 disturbance instead of a long train. In a layered crust, however, 

 we find a qualitative explanation in the effect of dispersion (i. e., 

 the dependence of the velocity of a wave on its period), which con- 

 verts an impulse into a series of approximately harmonic waves ; the 

 amplitude of Rayleigh waves dies down with depth in a distance 

 proportional to the wave length, so the short waves are confined to 

 the upper layer, in which the velocity is least, while the longer waves 

 extend into the lower layer where the velocity is greater. A quanti- 

 tative analysis of the Rayleigh waves is not at present available 

 because the variation of their velocity with wave length has not yet 

 been worked out. A further difference between the simple theory 

 and the observed facts is that the motion in the theoretical Rayleigh 

 wave is wholly in the plane containing the vertical and the direction 

 of propagation, whereas the actual motion in the long-wave phase 

 of an earthquake has a strong component at right angles to the 

 direction of propagation. This difficulty was removed in 1911 by 



