58 REPOKTS ON THE STATE OF SCIENCE. — 1916. 



S observations would show a rise : to get rid of the rise in them we 

 should have to add 30^°. There are recorded cases of the stoppage of 

 clocks near the epicentre which would be inconsistent with such large 

 corrections to the time at origin. 



On the whole, the case for the rise being real seems fairly strong. 

 And now we have to consider how to draw a smooth curve so that these 

 values shall be the means of groups. 



Suppose first we join the points by straight lines and let us further 

 omit the point for 13° and join 8° to 1S° by a straight line. The value 

 indicated for 13° would be J-(120s + 257s) = 189«. Now the observed 

 mean value 204^ lies 15= above this: and this is only the C.G. of the 

 triangle formed by the proper values for 8°, 13°, and 18°. The proper 

 value for the apex of the triangle would be at tlu'ee times the height; i.e., 

 30' above the C.G. Thus the proper value for 13°, interpolated 

 between 8° and 18° so as to make a triangle with C.G. at 204% would 

 be 234^. The points would then be 



8° 13° 18° 

 120 234 257 



Mean 5P 22'8 4-6 



"We see at once the necessity for a small value of SP following the 

 peak. Now doubtless the peak is not sharp but is rounded off; but 

 note that if we round it off we must at some point either increase the 

 large 2P = 22s-8 or decrease the small 8P = 4''"6; perhaps both. For 

 any process of rounding off the peak means that we must go outside the 

 triangle to make up the area lost from the peak. 



There is thus no difficulty at all about a small value of S P between 

 13° and 18°; indeed, it is almost a necessity. And hence the PX 

 phenomenon can probably be explained. The small value of SP 

 comes earlier than was suggest-ed in Table V. : but it seems probable 

 that by some little adjustment the phenomena may be all brought 

 into line. The reason why the sudden drop was assumed to come later 

 was the avoidance of the rise in SP near the epicentre. It seemed 

 theoretically probable that the velocity near the epicentre was nearly 

 constant, and thus, in order to accumulate a fund of positive errors 

 before the drop, SP had to be carried on at the highest available value 

 for some distance. Once the possibiHty of a rise in SP near the 

 epicentre is admitted and the drop may come earlier. But the initial 

 rise in SP is distinctly surprising, though the observations seem to 

 leave no room for doubt. 



