TERRESTRIAL MAGNETISM, ETC. 255 



as in a uniformly magnetized terella, but with its 

 axis thus oblique to the axis of rotation. If this 

 were exactly the case, the directions indicated by 

 the compass would lie along great circles passing 

 through the two magnetic poles, and the angles at 

 which these circles cut the geographical meridians 

 would be the actual variations in different parts of 

 the earth, and the magnetic equator would be a 

 circle on the earth's surface midway between the 

 magnetic poles, inclined to the astronomical equator 

 at an angle of 20. But, in fact, there are irregu- 

 larities of distribution, such as those adduced by 

 Gilbert to account for variation ; only we do not 

 find them related to distributions of land and 

 water, as he imagined. 



It is curious to find the idea of headlands attract- 

 ing the compass still cropping up again and again 

 two centuries after it was first suggested by Gilbert, 

 and fifty or one hundred years after advances in 

 knowledge of terrestrial magnetism had shown it 

 to be erroneous. I find in an unpublished letter 

 from the late Archibald Smith to Lord Cardwell, 

 of date 1 3th of February, 1866, which has been 



