ON SEISMOLOGICAL INVESTlGATtONS. 



4? 



correction of about 15^ to the time at origin is supported by these near 

 stations which would also (as will be shown in a moment) bring more 

 distant stations into better accord. The obsei'ved times being in excess, 

 the moment at origin must be altered from 8"^ 44"" 34^ to 8^^ 44°^ 49^ ; and 

 with this time at epicentre we get the column — Ca. It is clear that 

 Manila and Mizusawa cannot be brought into accord with the rest by 

 any change of epicentre, for the latter lies in nearly the same azimuth 

 as Tokyo and Osaka, while Manila is in nearly the same as Tsingtau 

 and Zi-ka-wei. 



Turning now to the results for stations more than 90° from the 

 epicentre, the Szirtes' curve as drawn suggests a curious phenomenon. 

 The slope has been nearly steady between 30° and 90° ; it then decreases, 

 especially between 100° and 150°, and finally increases ; the final slope 

 being at the rate of five minutes in 24° (or 12^'5 per degree, the same 

 as that at about A =18°. Hence if this were the correct cm've, we 

 should still have the phenomenon of anomalous reflection, though in a 

 different way. Two arcs, one of 18° and the other anything greater 

 than 108°, would combine to give a total path of 126° and upwards, 

 not because the value of 8P falls to 4* per degree at about 22° from 

 the epicentre, thus matching the small values at A = 100° onwards, 

 but because the value of 8P rises at 12^"5 at distances > 106°, thus 

 matching the large value at A =18°. But the correctness of this 

 interpretation is here challenged. Surely the rate SP diminishes to 

 zero at A = 180° ? It seems difficult to avoid the conception of a path 

 diametrically through the earth for A = 180° ; and paths lying near 

 this must be so nearly similar in all respects that the time to neigh- 

 bouring points must be nearly the same. Hence near A = 180° the 

 value of SP must tend to zero, as suggested in fig. 4; and if the 

 graph of SP rises in the manner indicated by Szirtes it will have 

 ultimately to come down again all the more. 



The interpretation now put upon the records at distances greater 

 than 105° from the epicentre is as follows: — 



(a) Pour or five are regular P waves, viz. : — 

 Table XXL 



The column — Ci is sensibly the same as Szirtes' results, and is got 

 with his time at origin and the Shide tables. In O-C2 the corrections 

 to Shide tables given in the last two reports are used, viz. : — 



A = 55° 65° 75° 85° 95° 105° 115° 



Corr" to P 



3 -( 



-15 -24 (-40) 

 The correction at 115° (not given before) is estimated from Table X. 



