4« 



REPORTS ON THE STATE OP SCIENCE. — 1916. 



Table X. 



But at the same time the differences for 130° and 140° are too large 

 to be passed over. It has been remarked in the last two Reports that 

 the tables for P and S seem to require sensible corrections at a distance 

 from the epicentre. For A = 105° the correction to time for P is given 

 as — 24', and is rapidly increasing : a correction of — 40' at A = 115° 

 is not out of the question ; and since the ' calculated ' result for 140° 

 depends on times for 25° + 115° the above large value of — C may 

 be chiefly due to the errors of adopted tables. In the column — C2 

 corrections to the tables have been applied. Here again we may get 

 help in coi'recting the tables by study of the reflected phenomena, 

 though direct observations of P are rare. 



As one more check let us turn to the record of the earthquake of 

 1913, March 14, which was very cai'efully worked up at the I.S.A. 

 Central Bureau by S. Szirtes (' Mitteilungen,' p. 117). His interpreta- 

 tion of the observations is shown by his diagram, here reproduced 

 (Fig. 5) with the addition of a rough network of lines and some larger 

 figures, those in the original being so small as to be scarcely legible. 

 A scale of degrees has further been substituted for that of kilometres. 

 (Is it not rather unfortunate that kilometres have been used so much? 

 There are many advantages in working with degrees.) For the present 

 we confine attention to the P curve. 



First of all let us see how the suggested new tables fit the observa- 

 tions near the origin. For this we turn to the figures given in the accom- 



Table XL 

 1913 March U^ 8^ 44™ 34». 3-5° N. 125-50 E. (Szirtes). 



Manila 



Batavia 



Taihoku 



Zi-ka-wei 



Osaka . 



Tsingtau 



Tokyo . 



Mizusawa 



Sydney 



Observed P 



120 

 210 

 220 

 28.0 

 32-4 

 330 

 34-8 

 38-4 

 446 



184 

 332 

 346 

 364 

 410 

 409 

 437 

 456 

 527 



panying text of Szirtes' paper and extract the following particulars. 

 The errors — Ci are with the tables in use; — Cj are with the 

 new tables above proposed. It will be seen that the new tables remove 

 a great part of the anomaly shown by Batavia and Taihoku, and that a 



