MAGNETIC SURVEY OF JAPAN FOR THE EPOCH 1895.0. 35 



+ 6' 46 in the Declination, 

 ±5:47 „ „ Dip, 

 + 73.' 2 ,, „ Horizontal Intensity. 



The probable errors of the empirical coefficients are not in strict 

 sense the result of accidental errors of observations, but are 

 chiefly due to the amount of local disturbances ; if we omit a 

 few of the stations which give large residuals, they will be 

 greatly reduced, or if we include some of the stations that were 

 omitted in the equations of condition they will be much in- 

 creased : they are put here simply to indicate some measure of 

 certainty in the values of these coefficients calculated by the 

 definite method, and also to show the danger of applying these 

 formulae to too large values of co-ordinates ; their extreme 

 amounts within the country are : — 



at 



\j<p=0 '^ [Jç) = + 8° 



±0' 68 + 1,5.' .5 in the Declination, 



±0: 58 ± 13: ,, ,, Dip, 



±8.' 1 ±183.'l „ ,, Horizontal Intensity. 



Their large increase with the co-ordinates is caused principally l)y 

 the smallness of weight in the coefficients of JÀJ<f, and they can 

 be diminished by taking simpler formuUe than the complete 

 quadratics, the eftect of increase in the weights of the co- 

 efficients over-compensating the increase of residuals. 



This is exemplified in the following reduction of the decli- 

 nation by the parabolic formula, made for the })urpose of finding 

 better value of the annual variation by comparing with the 

 expression of the same form obtained in the previous survey. 



