30 A. TANAKADATE. 



This method gives only the average secular variations, the 

 results are therefore partly over-corrected and partly under- 

 corrected especially in regions under the process of tectonic 

 change, some of \Yhich had already drawn our attention.* To 

 minimize such effects the nearest round number of years to the 

 mean time of all the observations is taken as the epoch to 

 which all the observations are reduced, the amount of maximum 

 correction l)einir about one and a half year either wav. 



It is to be remembered in this respect that three destructive 

 earthquakes have occurred between the previous survey and the 

 end of the present : the strongest in 1891 in the district of 

 Mino-Owari and two less severe, one in Sakata in the winter of 

 1893 and another in Tokyo in the summer of 1896. 



§ 8. Reduction to the Sea Level. 



To reduce all tlie observations to the sea level, the vertical 

 variations of the magnetic elements were derived in the way dis- 

 cussed in § 11 below, using the first approximation of mean 

 isomagnetics used in deducing annual variations. The corrections 

 are quite sensible in some of stations which are two or three 

 kilometers high and affect materially the amount of disturbing 

 forces in such altitudes. 



§ 9. Isomagnetics. 



The reduced values of magnetic elements were ^^ut on maps, 

 one for each element, and isomagnetics were drawn l)y the tent- 

 ative method of interpolation, taking care to give slight allow- 

 ances with respect to second differences. In some plac-es it was 



* The Disturbances of Isomagnetics atteudiiij? the Mino-Owari Eartliquake of ISUl, 

 A. Tanakadate and H. Nagaoka, Journal of Science College, Vol. V. part II. 



