270 REPORTS ON THE STATE OF SCIENCE, ETC. 



30 sec. after P is listed provisionally in the Kew Bulletin as PcP, and is 

 obvious on all three components. The records indicate another P between 

 these two, but it is less sharply defined ; I read this intermediate onset on 

 the Oxford and Dyce records as 10 and 15 sec. after the calculated times of P, 

 and the Kew Bulletin lists it at 15 sec. Only one station seems to have 

 suspected a multiple shock : Upsala interprets the readings as indicating 

 two aftershocks 20 and 33 sec. after the calculated time of the first shock, 

 and identifies the corresponding PP, S, etc. It is rather diflScult to pick out 

 S on the seismograms, and to avoid personal bias it is better to examine the 

 station readings. 



All the P readings in the I.S.S., including the ' Additional Readings,' 

 were analysed. The additional readings gave the following distribution for 

 residuals for P, from 9 to 42 sec. : 8 residuals above this were sparsely 

 scattered and could be ignored. 



26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 

 20241217622021 I I I 



The concentration of residuals about 33-34 sec. is unmistakable, especially 

 if, as a rough method of clearing out random large errors (Jeffreys, Geo- 

 physical Supplement, 2, 335), we make a uniform deduction of one per 

 group. There is rather slight evidence of onsets at about 12 and 21 sec. 

 after the calculated P. A corresponding analysis of all readings available 

 for S gave : 



o I 2 3 4 5 6 7 8 9 10 II 12 13 

 1222141753 s I 5 2 



14 IS 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 

 4017495380531253352441 



36 37 38 39 40 41 42 43 44 45 4^ 47 48 49 5° 

 131111211010011 



These S residuals show decided concentrations round about 10 sec, 

 19—20 sec. and 31 sec. For the first shock S seems to have escaped notice. 

 Taken all together, the evidence leaves no doubt that the shock is double, 

 and it is possibly treble or quadruple. Ideally, of course, a complete cor- 

 respondence should be observable among the onsets corresponding to each 

 phase ; actually, the phases other than P are not so immediately obvious 

 in the seismograms examined, and there is always the risk of finding just 

 those things that one expects to find. 



A check on the accuracy of the I.S.S. epicentre was made by taking the 

 residuals in three well-defined azimuth groups, the Japanese, European and 

 North American stations, and applying the correction for ellipticity recently 

 found by Bullen ; the shift indicated is less than one-tenth of a degree, and 

 is barely significant. 



It is worth noticing that the S residuals of the shock of 193 1 August 18^, 

 14*^. from the same epicentre are not abnormal, but form a rather ' flat ' 

 distribution round about — i, in which the only anomaly is a group of 

 12 residuals + S, where the general run would indicate about 4 such 

 residuals ; it is very doubtful if any significance should be attached to this 

 isolated group. 



