224 



circle was turned forwards and backwards, alternately, through 

 two circumferences, and the scale-readings noted after each 

 change. The following are the results : 



I. Magnet unloaded. 

 vi - Va = 720° ; Ml - W2 = 29'-31. 



II. Magnet loaded. 

 vi-V2= 720° ; Ml - M2 = 43'-43. 



III. Magnet unloaded. 



vi - ?;2 = 720° ; ui-U2 = 29'-22. 



Hence we have 



p = 1477 ; p = 995 ; and ^, = 1-484. 

 p 



Accordingly u = - 3'*28 ; and w = - 80° 45'. The values of 

 u deduced from the separate observations differ only by 

 0-04. 



In order to determine the effect of hygrometric changes on 

 the torsion of the suspension thread, a weaker magnet was 

 attached ; and the air within the box was alternately moistened 

 to saturation by wet sponges, and dried by chloride of calcium, 

 great care being taken to render the box air-tight. The po- 

 sition of the magnet was observed at an interval of some hours 

 after each change ; and the actual changes of declination were, 

 under these circumstances, eliminated by the help of simul- 

 taneous observations with an auxiliary apparatus. Finally, 

 the observed changes were reduced in the ratio of the magnetic 

 moments of the two magnets. The partial results presented 

 considerable discordance, notwithstanding every care in the 

 observations ; but the final mean is probably not remote from 

 the truth. It gave, — as the total effect upon the position 

 of the magnet, produced by the transition from complete dry- 

 ness to complete saturation, — a change of + I'-O, which cor- 

 responds to a change of position of the plane of detorsion of 

 24i degrees. The greatest change (from its mean state) in 



