172 MR. K. D. OLDHAM ON THE PROPAGATION OF 



Of these the two records of the earlier of the two Turkestan earthquakes of 1 5th 

 August, 1897, being 13 minutes in advance of the time of origin, may very probably be 

 attributed to some local shock. Three of the others are practically simultaneous 

 with the origin, and one, omitting the Tokio record of the Argentine earthquake, 

 about 7i minutes later than the origin. That is to say, the record begins from 

 5 to 1 5 minutes before the arrival of the condensational waves. 



It is very difficult to decide whether these early commencements of the record 

 have any real connection with the earthquakes they appear to refer to, or are due to 

 other, possibly local, disturbances which happened to coincide approximately with 

 the greater earthquake. 



On the one hand the number of cases in which there is an early commencement 

 of the record seems too great for the connection to be fortuitous. Excluding the 

 second Turkestan shock, where the record began at Ischia and Catania about 13 

 minutes before the earthquake, and the disturbance may well be attributed to some 

 other cause, we have no less than five out of ten distinct shocks, in which there 

 is a commencement of the record in advance of the disturbance of what I have 

 called first phase. 



On the other hand there is the want of accordance in the times, and the fact that 

 the early commencement was in each case only found at a single station ; as these 

 are about evenly divided between the light and heavy pendula, there is no guide as 

 to the nature of the disturbance. 



If due to the principal shock and not to local disturbances, these early commence- 

 ments of the record can hardly be attributed to any form of wave motion set up by, 

 and at the same time as, the earthquake. They would, in this case, have to be 

 attributed to premonitory disturbances of a nature very different to that of the 

 main shock, for, though unfelt in the neighbourhood of the origin, the initial energy 

 of the disturbance would have to be great enough to affect instruments at distances 

 ranging from one ninth to one quarter of the circumference of the earth. 



On the whole, then, it seems more natural to attribute these early commence- 

 ments, Avhich show no concordance in their times as compared with each other, to 

 local disturbances, or at any rate to some cause other than the earthquake with which 

 they are approximately coincident. A possible exception to this is the Tokio record 

 of the Argentine earthquake ; this, as suggested above, may be due to the earlier 

 emergence of condensational waves which have traversed the central core of the 

 earth, as compared with those which have not penetrated so deep and, though 

 traversing a shorter course, have done so at a lower rate of propagation. 



The results obtained in the preceding investigations may be summarized as 

 follows : 



1. The complete record of a distant earthquake shows three principal phases of 

 increase of displacement followed by decrease, the phases l>eing marked by a more 



