189S-1902. No.6.] TERRESTRIAL MAGNETISM. 33 



With regard to observations made at Station IV in the summer of 

 1902, the conditions are somewhat different. As already indicated, 

 there is every probability that the magnetic moment of both deflectors 

 was the same as during Amundsen's determinations of the constants at 

 Potsdam about the 1st December of that year. 



The value for log — , deduced by Amundsen's observations, may 

 ^' 



therefore straightway be employed, namely, 



1 ^ 



for magnet Y, 0.40318 

 for magnet 77, 0.39500 



On the other hand, there is a great difficulty to face when an en- 

 deavour is made to find a usable value for log (7^«, as, for the deflection 

 observations,' the new long rod was employed, and the double needle 

 P was always used as deflected magnet instead of the small needle L. 



Among Amundsen's series of observations from Potsdam, there is, 

 it is true, a set of deflections with magnet 7/ placed upon the new rod 

 at both the distances marked upon the latter, which I will call e' and 

 E'; but as in these observations the small needle L was used as de- 

 flected magnet, no use, unhappily, can be made of them. 



My only alternative has therefore been to seek for a mean value 

 for log Qt by the aid of the vibration observations made at Station IV 

 itself. 



Assuming that the horizontal intensity has remained unchanged 

 during a set of vibrations and a set of deflections, the following equa- 

 tion, by the combination of formulae (1) and (2), is obtained: 



n 

 log C^i = log 2 log r„ + log sin cp„ . 



(X 



On the 20th, 25th and 30th June, and the 1st July, 1902, there 

 were taken, from about 9 or 9.30 a. m. until late in the afternoon, first 

 a double series of vibrations with both magnets, then an entire set of 

 deflections with the magnets at both distances, and finally another double 

 series of vibrations with both magnets. 



We have thus obtained, for each of the above-named days, a mean 



value for T„ and for cp„, which may be assumed to correspond more or 



less with each other. 



n 



If these mean values, and the previously given value of log -, be 

 put into the above equation, the following values are obtained for Oft: 



