84 On the Annual and Semi-annual Seismic, Periods. [June 15, 



tude of the annual period ranges from O05 (New Zealand) to 0'67 

 (Sicily and Algeria), the average of 57 records being 0*33. 



2. Semi-annual Period. Of the 62 records examined, only 3 fail 

 to show a semi-annual period, the cause of the failure in these cases 

 being no doubt the imperfection of the seismic record. In New 

 Zealand and South-east Australia, the maximum epoch generally 

 falls either in February or March and August or September ; in 

 North America, as a rule, in March or April and September or 

 October. But for other regions it does not seem possible as yet to 

 deduce any law. The amplitude of the semi-annual period ranges 

 from 0'06 (southern hemisphere) to 0*79 (Mexico), the average value 

 being O24. 



3. In fifteen cases, the amplitude of the semi-annual period exceeds 

 that of the annual period. Eleven of these records include the 

 following insular districts, which are among the most well-marked 

 seismic regions in the world, namely, the Grecian Archipelago, 

 Japan, the Malay Archipelago, New Zealand, and the West Indies. 

 The average amplitude of the annual period in these eleven cases is 

 O16, and that of the semi-annual period O24; i.e., the average 

 amplitude of the annual period is just half that for all the districts 

 examined, while in the case of the semi-annual period the average 

 amplitudes are the same. 



Origin of the Annual Period. In this, the concluding, section of 

 the paper, an attempt is made to show that the annual change in baro- 

 metric pressure may be the cause of the annual change in seismic fre- 

 quency. It would be difficult to prove that such a connexion exists, but 

 reasons are given which seem to render it in some degree probable. 



1. The most probable cause of the origin of the majority of non- 

 volcanic earthquakes is the impulsive friction, due to slipping, of the 

 two rock-surfaces of a fault. Now, whatever be the causes of 

 seismic periodicity, it seems probable that they are merely auxiliary, 

 and determine the epoch when an earthquake shall take place, rather 

 than there shall be an earthquake at all. Professor Gr. H. Darwin 

 has shown that the vertical displacement of the earth's surface by 

 parallel waves of barometric elevation and depression is not incon- 

 siderable, and that it diminishes at first very slowly as the depth 

 increases. Since the fault-slip which produces even a moderately 

 strong shock must be very small, and since the work to be done in such 

 a case is, not the compression of solid rock, but the slight depression 

 of a fractured mass whose support is nearly, but not quite, withdrawn, 

 the annual range of barometric pressure does not seem incompetent 

 to produce the effects observed. 



2. Comparisons between the dates of the maximum epochs of the 

 seismic and barometric annual periods are made in 31 of the districts 

 treated in this paper. The seismic maximum approximately coincides 



