412 F. OMORT : 



The mean epochs given in the third cohimii of Table IX, 

 each of which indicates tlie mean of the dates of those eartli- 

 quakes included in one group, may be interpreted as represent- 

 ing the epochs of greatest activity of destructive shocks. Again, 

 the differe7ices given in the fourth column of tlie same table, 

 which are the intervals between the successive mean epochs, vary 

 between 7 and 23 years, the average value being 13:\ years. Tlie 

 conclusion is that the epochs of maximum activity of destructive 

 shocks would recur on the average in every 13 or 14 years. 



14. It is impossible to determine whether there was in 

 historic times a permanent variation in the frequency of des- 

 tructive earthquakes in Japan as a whole ; the increase of earth- 

 quake records in later centuries implying, of course, not neces- 

 sarily any such variation. The frequency seems, however, to be 

 subject to certain fluctuations of long period. Thus the curve in 

 fig. 6, which represents the variation of the 10-yearly activity 

 of destructive earthquakes in Japan, indicates a maximum in 

 the middle of the 17th centur}^ and one minimum in the middle 

 of the 16th and another in the middle of the l<Sth century, the 

 length of the period being in this case about 200 years. Again, 

 according to fig. 7, which shoAvs for the whole country the varia- 

 tion of the activity of destructive earthquakes in every half century, 

 since the latter half of the (îth century (see Table XII), there 

 seems to exist, in the interval of about 1000 years between the 

 middle of the 8th century and that of the 18th a more or less 

 definite series of "> maxima and minima, giving a period of an 

 average length of 200 years. 



15. To see whether the long-period fluctuations of ordinary 

 small shocks and destructive earthquakes take place simultane- 

 ously, I shall compare the curve of the half-century activity 



