224 REPORT— 1898. 



Inasmuch as these records have been obtained from different types of 

 instruments which have had different degrees of sensibiUty, it is clear 

 that they cannot be regarded as individually comparable ; but when 

 plotted on squared paper and taken in groups, it is evident that these 

 time intervals increase with the length or depth of the wave-path over or 

 at which a disturbance has travelled. In a few instances, as for example 

 in the case of disturbances originating near Japan, Borneo, and Hayti, 

 which have been recorded by the same or similar instruments in the Isle 

 of Wight and Toronto, such observations are comparable, and they take up 

 expected positions on the average curve of duration drawn through the 

 groups of observations which are not so strictly comparable. 



The expectation from this is that this curve will, by future observa- 

 tions, be found to be approximately correct. 



An inspection of the same shows that the preliminary ti'emors up to 

 distances of 12° or 15° only outrace the succeeding waves by intervals 

 seldom reaching a minute. On paths between 20° and 85° the intervals 

 ai-e proportional to the length of the arc, but beyond this range it seems 

 that they may increase at a somewhat higher rate. 



Between Europe and Japan, or a distance of 85°, observations have 

 shown that the interval by which the larger waves are outraced varies 

 from 30 to 34 minutes. If we take 32 minutes as an average, then it is 

 easy to compare what should be expected, and what has been observed on 

 ranges lying between 20° and about 100°. This is done in the following 

 table : — , 



Japan to Shide, 85° Observation, 32 minutes. 

 Borneo to Shide, 112°, 42 min. expected. „ 40 to 43 „ 



Haj'ti to Shide, 62°, 23 „ „ „ 20 



Hayti to Toronto, 20°, 7 „ „ „ about 4 „ 



Although these observations indicate a working rule, enabling us to 

 determine the distance of an origin from an observing station which, with 

 a knowledge of the surface configuration of our globe and localities where 

 seismic activity is frequent often, are the means of locating an epicentre, 

 the last of the series suggests that the duration of the preliminary tremors 

 are more directly connected with the depth of a wave-path rather than its 

 length, as represented by the arc of a great circle. 



Trial, however, shows that the duration of preliminary tremors is not 

 proportional to the length of the chord along which it may be supposed 

 the movements travelled, or to its maximum or average depth. 



The table on p. 221, which is derived from fig. 4, shows that the dura- 

 tion of preliminary tremors in minutes is, for the given ranges, nearly 

 equal to the square root of the average depth of the chord expressed in 

 kilometres. 



On the Period of Earthquake Waves at Great Distances from their Origin. 



All that we know about the period of earthquake waves after they 

 have travelled great distances is derived from the open diagrams of the 

 Italian workers, a few records obtained in the Isle of Wight, and a single 

 but exceedingly valuable record obtained by Dr. F. Omori when in Pots- 

 dam. The Italian and Isle of Wight records were obtained from simple 

 or horizontal pendulums writing on smoked paper. The Potsdam record, 

 which refers to an earthquake originating in Japan on February 19, 1897, 



