utilizing- the same data of monthly distribution. I divide the 

 year into seasons relative to the earth's equinoctial periodis, 

 and the intervening solstices. This, of course, suggests a line 

 of causation not considered by Mr. Taber in his paper, and 

 involves a theory not yet established. Still it is worthy of 

 investigation on a strictly physical basis such as I propose, 

 and if periodicity is indicated empirically, we may bring for- 

 ward the theoretical part of the discussion at leisure. I have 

 merely to suggest at the outset that, the equinoxes them- 

 selves being periodical, if we succeed in allying earthquakes 

 with them in any degree as effects or as concurrent phe- 

 nomena, we shall have gone a long way toward solving the 

 standing problem of seismic periodicity. 



The vernal equinox occurs March 21st. Allowing that 

 there may be some relationship between this occurrence and 

 seismic action, it is fair to assume that the equinoctial influ- 

 ence begins some time before the actual event, and continues 

 some time afterward. Let us assume, for convenience, that 

 the period of greater or less equinoctial influence is three 

 months, beginning February 1st and ending April 30th. This 

 would allow 48 or 49 days before the equinox and 40 days 

 after it. Granting this amplitude for the season of the vernal 

 equinox, we should surely find indications of seismic increase 

 or decrease within it if there be such a tendency developed in 

 the curves. 



In a similar manner we set off the months of August, 

 September and October as the season of the autumnal equi- 

 nox. The intervening seasons of three months each are 

 allotted to the summer and winter solstices, and we have the 

 year divided as follows : 



Season of vernal equinox — February, March, April. 



Season of summer solstice — May, June, July. 



Season of autumnal equinox — August, September, Octo- 



Season of winter solstice — November, December, Janu- 



ber. 



ary. 



When we come to consider the year thus divided with 

 respect to seismic frequency, the only striking variation from 

 the plan usually in vogue is that the year begins with Feb- 

 ruary, and January comes trailing along after December, as 

 it in fact always does in the natural round of years. The 

 advantage of this readjustment is simply to bring out in our 

 diagrams the equinoctial and solstitial eft'ects, if they exist, 

 in stronger relief. 



Re-drawing Mr. Taber's most typical curve from his 

 plate 7 (that of total South Carolina earthquakes during the 

 years 1886-1913), we have the following: 



39 



