252 REPORTS ON THE STATE OF SCIENCE, ETC. 



although in the case of some of the smaller shocks it may be that more 

 grouping could have been effected. However, the determinations have 

 been made separately, and only those which prove themselves to be repeti- 

 tions have been adopted as such. 



Another series of shocks occurred on 193 1 November 1 and 2 off Shikoku 

 Island, Japan. There were three large shocks, but the waves of the smaller 

 shocks are difficult to allocate to definite epicentres and have been listed 

 under the stations which recorded them. 



Work on the data for March 1932 is now in hand, but a number of 

 stations have not yet sent in reports for that year. This is much to be 

 regretted. Even if the reports arrive in time for the observations to be 

 inserted before the copy goes to press the dilatory stations do not pull their 

 weight in the determinations of the epicentres and times of the earthquakes. 



Transmission Times. 



By Dr. Harold Jeffreys. 



The revision of the tables mentioned in the last report has now been 

 published by the Bureau Central de Seismologie. In later work based on 

 the same data, supplemented by estimates of the thicknesses of the upper 

 layers from deep-focus earthquakes and surface waves, and by earthquakes 

 well observed at short distances, I have obtained a formal solution for the 

 times of P so long as it does not cross the 20 discontinuity, and by combining 

 this with the times beyond 20 I have found an estimate of the depth of the 

 discontinuity, which is 483 ± 17 km. below the outer surface, 42 km. of 

 this representing the adopted thickness of the upper layers. Times of P 

 have been calculated for focal depths down to the discontinuity. 



The work is now being extended to S and SKS. The difficulty about 

 these pulses is that about 20 and beyond 70° the residuals do not fit the 

 normal law of errors even approximately, and the correct method of treat- 

 ment is uncertain. At these distances various published tables differ by 

 10 or 15 sees. A test has been obtained from the deep-focus earthquakes 

 discussed by Scrase and Stechschulte, additional Japanese observations 

 published by Wadati being used in both cases. These give satisfactory 

 series of S observations for rays that have not crossed the discontinuity, 

 and show that up to 20 the times in the Jeffreys-Bullen Tables can be 

 trusted to about 2s. To convert into actual travel times from a surface 

 focus the times of P need to be increased by about 9s., and those of S by 

 about 14s. Beyond 70 , however, a substantial decrease of the times of S 

 (with respect to those up to 20 ) is indicated by these deep-focus earth- 

 quakes and a number of normal ones ; the same applies to SKS, the 

 difference reaching about 10s. Other material is being incorporated, but a 

 satisfactory separation of the various movements that are read as S cannot 

 be obtained unless the epicentre can be fixed with a standard error of o- 1° 

 or so ; and it is not often that a suitable epicentre is associated with a good 

 series of S observations. 



Times of pV, sP, sS, and sSKS have been calculated ; comparison with 

 observation, however, suggests that the above estimate of 42 km. for the 

 total thickness of the upper layers is about 6 km. too great. 



The rise in the velocity of P or S at the 20 discontinuity is about 9 per 

 cent. ; Bullen, using the theory of the figure of the earth, finds that an 

 increase of about 10 per cent, in density is also necessary. A suitable 

 material to agree with these values is hard to find, but Dr. J. D. Bernal has 

 suggested an explanation based on the properties of magnesium germanate, 



