ON THE AFTER-SHOCKS OF EARTHQUAKES. ]^29 



for Gifu (Figs. 7, 8, and 9) indicate respectively fiuctuations, whose 

 average length are 4|^, 12.3, and .33 days. Longer periods are not 

 evident. 



The curve of monthly earthquake frequency for Gifu (Fig. 5) 

 indicates four maxima and minima between November, 1891 and 

 December, 1893, whose dates are as follows. — ■ 



Maximum. Miniumm. 



April, 1892, June, 1892, 



September, ,, F'ebruary, 1893, 



April, 1893, June, ,, 



September, ,, December, ,, 



The intervals between successive maxima or minima are from 4 

 to 8 months. 



§ 22. The Kagoshima Earlluiuahc. 



The curves of daily and 2-daily earthc^uake frequencies for Cliiran 

 (Figs. IG and 17) indicate respectively periods whose a^erage lengths 

 are 4.4 and 12 dnys. The curve of 5 -daily earthquake frequency 

 (Fig. 18) seems to indicate a period whose average length is about 

 33 days. 



The Kagoshima eartli(]uake is not yet sufiicientlv old to give 

 indications of longer periods. Tlie after-shocks of this and of the 

 Kumamoto earthquake were not numerous enough to enable us to 

 draw curves of hourly frequencies. 



§ 23. We have before noted that a few severe after-shocks are 

 likely to be followed by their own after-shocks. It might be sup- 

 posed that the maxima, which occur in the curves of monthly eartli- 

 <|uake frequencies (Figs. 1 and 5) represent merely the effects of such 

 shocks and not the real fluctuations in the residual elfect of the initial 

 earthquake itself ]]ut the fact is, on the contrary, that strong shocks 



