MAGNETIC SURVEY OF JAPAN FOR THE EPOCH 189Ö.0. 23 



the times of observations in the two surveys : these fell respec- 

 tively 1887.51 and 1895.12 during whicli interval the variation 

 is assumed to have ])een uniform ; thus : — 



d=:)° o.i'(;4-o.'2S6'Jç'-o:i44j;.-o:ooo294(j;.)- at 1895.12 



0=4° 54.'4+0:241J^-0.'l33J;-0.'000231(J/)- „ 1887.61 



J.;= li:2 + 0:04.3J(P-0:011J>i-0:OOOOG3(j;0' for 7..3I year«. 

 and -'^-1:49 + 0:OO.j!J:JJ^--0:û314GJ/-0:0000084(J;0' per annum 



where J^" = (^— 37°)', J/^. =r(/ — 138°/ * expressed in minutes 

 similarly 



^ = .30° 47:4+i:0G8J^-0:0792j;. at 1895.12 



H=ho° .54:9 + 1. 'i4(;j^- on .5.5GJ;. ,, iss7.gi 



Jd=^ - 7:.3-0.'Ü78Jv^ + 0:07G4J/. for 7..'31 years. 



^^z= -.l.'00-0:0104Jc? + 0.'0l02J/ per annum 



11=29317- G'i2_f^-l!48j;. at 1895.12 



H = 29247'-G.'l7J^-l'.17j;. „ 18S7.G1 



JH= 7d + 0.'05Jy^-0'.31j;> for 7.51 years. 



^^= 9'.33 + 0'.00G7J^-0!041J;. per annum. 



The vahies for 1895.12 were derived from the results of 

 observations at 288 stations which w^re seemingly free from ex- 

 travagant local disturbances by the method of least squares, a 

 process somewhat superfluous but useful in training the com- 

 puters for further work. Tables II, III, IV, give these annual 

 variations for each degree of longitude and latitude throughout the 

 country and were used in reducing the values to the epoch 1895.0. 



* The origin is taken at round number of degrees instead of the mean value for 

 facilitating the use of the formuhe, tlie slight increase of probable errors in the comiuted 

 values thus caused is quite insignificant. 



