258 



REPORTS ON THE STATE OF SCIENCE, ETC. 



by dividing the intei'val between the last (standard) date in July and the first 

 in August by 21m. We thus have : — 



Diff. 516 x21m + 1-6 



The small difference of -} 1.6m. indicates a slight correction to the period. 

 Spreading it over the interval between the means of the groups, which is about 

 870 periods, the correction indicated is about 1.6/870=. 00134m., comparable 

 with what we shall find below. 



The Periodicity not Local. 



Tne question thus arose whether this periodicity was confined to earthquakes 

 from the same focus or was wider in scope. The July series above is of a few 

 days only and might conceivably be a local phenomenon (in space and time), 

 but the August series is scarcely in keeping with this view. If the earth 

 generally is affected, we must find the space relationship at any particular 

 moment. A few trials of shocks at about the same time in different localities 

 suggested that the longitude is unimportant and that the effect spreads from 

 the equator outwards to the poles. The rate of travel suggested that the journey 

 from equator to either pole is completed in 21 minutes, but the first approxima- 

 tion was found to require sensible correction, the rate being certainly slower 

 near the equator. Ultimately the following (quite provisional) table was found 

 to give good results : — 



Times for Different Latitudes. 



Lat. 



Time 



The information about very high latitudes is scanty, and 21m. may be set 

 down for the poles without much fear of contradiction at present. 



When all the earthquakes of 1917 were discussed in this way a distinct 

 periodicity in 21m. was manifest, but with some curious features requiring 

 further examination. After a good deal of work, and the use of all the large 

 earthquakes from 1913-16, the following features of the earth movement 

 emerged : — ■ 



(a) There is a distinct oscillation of the maximum in six months, the solstices 

 and equinoxes being the epochs of extreme range, which is about 120°, or a 

 one-third-period of 7 min. 



(6) There is no sensible annual oscillation, but 



(c) There is another oscillation, about 180° in amplitude (or 10.5 min.), in 

 twenty-four months, with spring equinoxes as extreme epochs. To so unexpected 

 a feature it is difficult at present to assign a meaning, but the very difficulty 

 lends some support to its reality. To show the kind of evidence for it the 

 following figures may be given. After removal of the six-month periodicity (a), 

 the results were collected in groups of six months (which would render any 

 still uncorrected six-monthly term insensible), and the maximum computed by 

 harmonic analysis. The phases of the maxima came out as follows for the nine 

 sets (of six months each) available, viz. two each in the years 1913, 1914, 1915. 

 and 1916, and one in 1917 which is not yet fully reduced. They are placed in 

 sets of four to show the 24-monthlv term : — 



Years 



Maximum Phases 



