176 KNOTT AND TANAKADATE 



Mr. Tanakadate worked alone by the usual eye and ear method. 



The object aimed at being to obtain a general magnetic survey 

 of Japan, it was advisable as far as possible to eliminate local and 

 diurnal disturbances. The stations had all orioinallv been chosen 

 after a careful study of the distribution of volcanoes throug-hout the 

 country. If further we assume that all mountains are a possible 

 source of disturbance — of underoTound sources of disturbance it is 

 impossible of course to take any preliminary account — the following- 

 rough rule, which guided us in most cases, v>qll ])robably be found 

 useful : — A station should be so placed that no mountain as seen 

 from it shall subtend a ^'ertical visual aniile greater than 5°. Thus, 

 let there be a mass of magnetic substances of volume v and suscep- 

 tibility k at a distance r from the station ; and suppose that this mass 

 is a cube* of height Jt. Then, hi be the intensity of the magnetic 

 held in which the mass is placed, the disturbance at the station will be 

 klv/r = kl(h/ry==k I tan'^ff, where is the visual angle. Take 

 7 = ••!, A; =10 (bad iron), = 5°, and the value of the disturbance 

 comes out '0027. In the most favoured circumstances, however, it 

 is highly improbable that more than 1 or 2 p. c. of iron is present in 

 the substance of the mountain. Hence we may safely assume that 

 a mountain at such a distance will only affect the 5th decimal place. 

 In a few cases, such as Wakwan (No. 62) Hagi (No. 72) Hamada 

 (No. 73), we were compelled by circumstances to break the above 

 rule ; but in no case did the visual angle amount to 10'\ 



The station was usually chosen in an open Held within 1 or 2 

 kilometres of some village or town. Occasionally a cotton field was 

 cleared sufficiently to permit the tent to be pitched. 



On a few occasions dip observations were made at two or three 

 spots in the vicinity of the chosen site ; and if the values came out 



* Such an assumption gives of course a higher value than is likely to be. [C. G. K.] 



