SEISMOLOGICAL INVESTIGATIONS 253 



which is chemically very similar to olivine. At ordinary pressures the 

 germanate exists in two forms found by Goldschmidt, a rhombic one 

 analogous to olivine, and a cubic one analogous to spinel, the latter being 

 the denser and therefore likely to predominate at higher pressures. The 

 silicon atom, being smaller than that of germanium, will interfere with the 

 further compression of the oxygen lattice of olivine at a higher pressure, 

 but the next stage can be inferred by analogy, and it appears that the material 

 between the 20 discontinuity and the core is likely to be a cubic form of 

 olivine. 



K. E. Bullen, following up his work on the density, is calculating the 

 effect of the ellipticity on the times of transmission. It is smaller if geo- 

 centric latitudes are used instead of geographical ones, the difference in 

 extreme cases reaching about 2s. for P ; apart from this the effect never 

 reaches is. The whole effect does not exceed o-4s. up to 30 . When this 

 work is complete it will be necessary to correct the present tables to adapt 

 them to a spherical earth, but this will not be difficult. 



A comparison of the accuracies of seismological stations has been carried 

 out by means of the P residuals for the best observed earthquakes in the 

 I.S.S. from January 1930 to March 1931. The bulk of the best stations 

 appear to attain a standard error in routine observation of 2s. or a little over. 

 This accuracy is reached in Great Britain only by Kew and Oxford. Some 

 of the apparent standard error is due to errors in the epicentres, but not 

 much ; most of the I.S.S. epicentres indicated by the marks N. 1 and 

 R. 1 appear now to be accurate to 0-2° or less, the probable errors as given 

 being too high. In some earthquakes, however, average standard errors 

 at all stations as low as 1 -3s. have been found ; I think that this is due to 

 special clearness of these shocks in comparison with others. In the North 

 Sea earthquake of 193 1 June 7, for instance, I have redetermined the epicentre 

 from the I.S.S. data, obtaining 53°-95 ±o°-os N., i°-55 ± o°-o6 E. ; 

 this makes the P residuals at five of the eight British stations equal to o or 

 ± is., two equal to + 2s., and one equal to —3s. Thus they can attain 

 high accuracy in favourable conditions. The general comparison, however, 

 is useful in selecting stations for a preliminary determination of an epicentre 

 and in adjusting the weights of doubtful observations ; for bad observations 

 can occur at even the best stations. 



Very Long Seismic Waves. 

 An Editorial Note. 



In examining the records • of the great submarine earthquake which 

 occurred in the South Pacific (long. 156 E., lat. 57 S.) on June 26, 1924, 

 W. C. Repetti and J. B. Macelwane noticed certain long waves which they 

 denoted by X and U. The X wave had an enormous amplitude at the 

 nearest stations for which the records were available. At Wellington, 19 

 from the epicentre, there was an oscillation taking 18 mins. At Sydney, 

 23 ° from the epicentre, the oscillation took about 9 mins. ; the amplitude on 

 the Wiechert seismograph was several centimetres. At Uccle, 162° from 

 the epicentre, there were trains of about seven waves with a period of a 

 minute. These waves, which had an amplitude of less than a millimetre 

 on the Galitzin records, were judged to have passed 1$ and z\ times round 

 the globe. Repetti found for the velocity of the X waves 4-51 km. per 



1 W. C. Repetti, S.J., Bull. Seism. Soc. Amer., 17 (1927), 207 ; J. B. Macelwane, 

 S.J., Gerlands B.z. Geophysik, 28 (1930), 165. 



