ON SEISMOLOGICAL INVESTIGATION. 75 



cases, except the case of Mauritius. If there is no mistake in calculating 

 the times, then the disturbance travels comparatively slowly along the 

 Alaskan Batavian route. This route, if it lies near the si(,rface, is almost 

 wholly beneath the deeps of the North Pacific. But then, on the other 

 hand, the Alaskan Mauritius route is also a comparatively slow route, 

 and it lies further to the west, under Siberia, India, and the Indian 

 Ocean. Still, these two routes are in the same quarter of the 

 globe, so that a similar value for the speed is not unlikely. It may 

 be not merely a question as to whether sea or land is overhead, but may 

 depend on the general character of the rocky material. These two routes 

 left out of accouiit, there is a very striking constancy in the value of the 

 arcual speed calculated for these various routes. In the four routes to 

 Shide, San Fernando, Bombay, and Cape of Good Hope, the great circles 

 pass all very near the poles. It is beautiful to see how well these four 

 polar routes agree, With the somewhat scanty material you have to 

 hand, I doubt if you would be at all warranted in making any deductions 

 as to variations of speed. The Alaskan results suggest a constant value 

 for the arcual speed. The same constancy is indicated in the Mexican 

 earthquakes, but the value comes out distinctly smaller than in the Alaskan 

 quakes. Why is this 1. Still thinking of great-circle routes, we see that 

 there cannot be much difference between the Mexican Batavian and the 

 polar routes from Alaska, unless, of course, the former goes pi'eferably by 

 way of the South Pole. But that possibility is not considered in calcu- 

 lating the speeds. If we took it that way the speed would come out 

 larger in the ratio of 210 to 150 or 7 '5, giving 1-9 instead of 1*4, a 

 remarkable coincidence truly. The Mauritius number will also be 

 increased in much the same ratio. But what are we to make of the 

 others ? No, I think we must get at an explanation of the much smaller 

 speeds associated with the Mexican earthquakes in some other way. Is 

 it possible that the depth of the seismic focus might have something to 

 do with it ? Have you any facts to guide you to an estimate of the 

 probable depth ? 



And now pass on to the Ceram quake. Here the constancy, so marked 

 a feature in the other cases, no longer holds. There is an undoubted in- 

 crease in the arcual speed over the longer arcs. The most striking feature 

 is the smallness of the Mauritius route speed as compared with that 

 associated with the Cape of Good Hope route ; for there cannot be much 

 difference in the routes for the greater part of the way. But did not 

 Mauritius give a too small value in the Alaskan earthquake also 1 Again, 

 I ask, is there no possibility of an error in the time estimate 1 Ceram 

 Victoria and Mexico Batavia give approximately the same value for the 

 arcual speed — a point which tells in favour of the accuracy of the time 

 estimates, for the routes are very different in the two cases. Leaving out 

 of account all but the broad features, we may conclude that the speeds 

 (arcual) associated with the Alaskan are distinctly greater than those 

 associated with the Mexican and Ceram earthquakes- But I confess I 

 can give no satisfactory explanation of this, nor can I see why Batavia 

 and Mauritius should give smaller values than the others in the Alaskan 

 group, and why Cape of Good Hope and Shide should give comparatively 

 large values in the Ceram group. 



And now let us see what comes of taking the chord as the approximate 

 path of shortest time. Interpreted in this way the results indicate that 

 the waves must go diametrically through the earth at a much slower average 



