SEISMOLOGICAL INVESTIGATIONS 271 



Seismic Transmission Times. 

 By Dr. Harold Jeffreys, F.R.S. 



The most important contribution of the year to theoretical seismology 

 is probably Bullen's determination of the systematic errors due to the neglect 

 of the ellipticity of the earth. The total effect on the arrival time of P at 

 large distances may vary with azimuth by about 5 sees. I have now applied 

 the corrections to the data used by BuUen and myself in the paper we sub- 

 mitted to the Conference at Lisbon in 1933 and to more recent earthquakes 

 suitable for determining the times of S and SKS. Times of P, S and SKS 

 (to 115°) adapted to a spherical earth are now available. Some systematic 

 errors were detected in the process. The maximum departure of the P times 

 from those of Gutenberg and Richter is about 2 sees. The correction intro- 

 duces a systematic correction into the epicentres, which is small (usually under 

 o°-05) for earthquakes in the northern hemisphere, but may reach o°-5 for 

 southern epicentres if near stations are few or receive insufficient weight : 

 27 southern epicentres have been redetermined, as a preliminary to a new 

 study of the core waves. It appears that errors in the epicentres may 

 account for the low reliabilities previously found for some of the southern 

 stations. The negative residuals at large distances, noticed by Bullen and 

 me in some southern earthquakes, disappear when the ellipticity is taken into 

 account, and there is no longer any need to suppose that Pacific earthquakes 

 habitually have a small but appreciable focal depth. A preliminary classi- 

 fication of the P residuals by distance suggests that the time to 50° is about 

 I -5 sec. shorter in Pacific than in continental earthquakes, but further ex- 

 amination will be needed before it can be said that this is not due to a small 

 systematic error of observation. A difference between oceanic and con- 

 tinental travel times might be expected on thermal grounds ; indeed, it is 

 rather surprising that the difference should be so small, corresponding to 

 about I part in 300. 



The seismological evidence indicates rapid changes with depth in the 

 velocities of P and S, possibly a discontinuity, at a depth of about 470 km. 

 Bullen, using the theory of the figure of the earth, finds that the mean 

 density down to the core is too high to agree with that of olivine in its normal 

 state and at the actual pressures. I referred in last year's report to Bernal's 

 suggestion that at high pressure the usual rhombic form of olivine may be 

 replaced by a cubic one of higher density. This hypothesis has been tested 

 by comparison with conditions in the moon. The relevant pressure is not 

 reached in the moon, and the density of the moon is nearly that of surface 

 olivine. Thus the density would agree with Bernal's hypothesis. It would 

 not agree, however, with the presence of a large amount of a different 

 material denser at normal pressures. The former hypothesis being there- 

 fore adopted, the moon can be shown to be nearly homogeneous. This 

 provides the additional equation needed to determine the earth's ellipticity 

 from the moon's perturbations, and gives i/e = 297 '2 ± 0-5, the accuracy 

 of which is comparable with that of the determinations from gravity, and has 

 the advantage that the doubtful effect of higher harmonics in the earth's 

 gravitational field does not arise. Dr. H. S. Jones, using slightly different 

 data, gets 296 08 ± 095. Prof. E. W. Brown and his collaborators are 

 likely to improve these values soon, as the method appears promising. 



Reappointment of the Committee. 



The Committee asks to be reappointed and for the renewal of the grant of 

 £100 from the Caird Fund. 



