138 MR. R. D. OLDHAM ON THE PROPAGATION OF 



treat the matter anew. With the exception of some studies of the rate of propa- 

 gation of earthquakes within the seismic area, the published calculations deal, almost 

 without exception, with isolated observations, not infrequently with earthquakes 

 whose place or time of origin are more or less uncertain, and seldom take any notice 

 of more than the time of commencement of the record, and in some cases also of the 

 maximum displacement. 



The very laborious series of researches by Dr. AGAMENNONE I have been unable to 

 make use of, owing to the method of calculation adopted by him. The time, accu- 

 rately determined at some place away from the origin, is taken, and the distances 

 from the origin of that place, and of the more distant places whence observations 

 have been obtained, are calculated ; thence the apparent surface velocity is obtained 

 by dividing the difference of distance by the difference of time. This method would 

 be perfectly legitimate if we were dealing with only one class of wave motion, 

 propagated at a uniform rate round the surface of the earth ; if, on the other 

 hand, the wave motion is propagated along a brachistochronic path through the 

 earth, it must necessarily lead to misleading results, owing to the difference between 

 the true and apparent rates of propagation, a difference which varies with the 

 distance from the origin. 



The object of this study being to determine whether the three phases recognised 

 in the records of the great earthquake of 1897 are a constant or an accidental 

 feature, as well as to ascertain the forms of wave motion and wave-path represented 

 by each phase, it is essential that the time intervals should be referred to the time 

 of origin, and not of arrival of the shock at some place away from it. It is also 

 necessary that these time intervals, as well as the distances from the place of origin, 

 should be determined with a close degree of accuracy, as without this it would be 

 impossible to decide whether an apparent resemblance in the records of different 

 earthquakes was, or was not, due to the same cause in each case. 



To ensure this accuracy, it is necessary to select the earthquakes dealt with. In 

 t\iQ first place, it is essential that the disturbance should originate in a single effort 

 of short duration. Fortunately all those which fulfil the other conditions fulfil this 

 also, and none have had to be rejected on this score. Secondly, the time and place 

 of origin must be tolerably accurately known. The limits of error adopted have 

 been 1 minute of time and 1 of arc. In the case of all the earthquakes noticed 

 below, these limits are not exceeded, and in several cases the limit of error from this 

 cause is much less. Thirdly, I have excluded all cases where there were not a 

 sufficient number of independent records to serve as a check on, and confirmation of, 

 each other. Fourthly, as a separation of the condensational and distortional plane 

 waves is not to be looked for in the heterogeneous materials near the surface of the 

 earth, the records from places at distances of less than 20 of arc from the origin 

 have not been taken into consideration. 



In collecting the facts I have been almost exclusively indebted to the careful 



