MAGNETIC SUKVEY OF JAPAN FOU THE EPOCH 1S9Ô.0. 129 



those empirical coefficients by imposing the condition of irrotation- 

 ality among them, as was suggested by the writer on previous 

 occasion. This being premissed, the vahies along the line of no 

 current will be represented nearer to the truth than the rest 

 and should be taken in preference to values at other places in 

 deducing the magnetic constants of tlie wliole globe. 



§ 12. Vertical Variations of Magnetic Elements. 



The variations of magnetic elements due to difterence of 

 level has recently l)een computed l\v Prof. I><iznar from 205 observa- 

 tions taken at different elevations during the magnetic survey of 

 Austria and Hungary. He arrives at results which are more 

 than three times as great as those ol^tained by taking the dif- 

 ferential coefficients of the first term in the spherical harmonic 

 expansion. This discrepancy is ascrilied by the author to prob- 

 able causes external to the eaith and doubt is thrown upon the 

 Gaussian method of rejiresenting the magnetic potential of the 

 earth. 



Later still, van Eijckevorsel and van Bemmelen made special 

 investigations on the subject I)y taking a large numl)er of obser- 

 vations on the Rigi and came to the conclusion that the varia- 

 tion of dip due to height are so small that they are almost 

 entirely masked by the instrumental and ol3servational errors. 



The chief difficulty in such case lies evidently in the fact 

 that observations at high level are made on mountains which 

 are themselves more or less magnetic presenting local disturbances 

 often exceedinof the mere effect due to elevation. 



The plan here adopted is to deduce those vertical variations 

 from their values observed on level surface, supposing the electric 



