to converge at the poles ; notwithstanding their deflexions and 

 undulations from various local disturbances, yet they preserve a 

 remarkable degree of regularity. Numerous fruitless investiga- 

 tions have been made with the view of ascertaining the position 

 of the magnetic pole, as if that must be a mathematical point. 



Any person who has had some experience with the action of 

 fluids, whether water, air, or electric, converging towards or di- 

 verging from a central passage, must know that they cannot be 

 forced into a mathematical point, there must be a limit to their 

 compression or density ; nor can it be expected that every indi- 

 vidual current should retain its exact radial course towards the 

 focus, and much less when diverging from it. Hence, from 

 analogy and observations, the narrowest limits that we can assign 

 to the polar axis to which the magnetic currents converge and 

 diverge are, perhaps, the areas bounded by the arctic and an- 

 tarctic circles. 



By taking these extensive spaces for the magnetic poles, we 

 shall be relieved from entering into the useless and endless in- 

 quiries respecting late discoveries, and formulas which have been 

 established or founded on them. The question is rendered 

 simple, not requiring formulae which profess great accuracy in 

 points where the data of observation must be very uncertain. 

 Although on an average the magnetic meridians, described in 

 Plate IV., may be considered to approximate to the average di- 

 rection at the present time, yet it must be k ept in mind that a 

 great proportion of the lines have been assumed ; however, they 

 are sufficiently near for our present question. 



Numerous observations have been made in the equatorial 

 regions, indicating both east and west variations in the same 

 meridian of only a few leagues in extent ; and numerous other 

 experiments and observations may be quoted to prove that the 

 direction of the needle does not necessarily point towards the 

 centre of convergence of the individual current which moves it, 

 but in the direction of the resultant, viz. the compound of the 

 primary and local currents — the diagonal of the parallelo- 

 gram of the two actions. The local disturbing force being 

 a variable quantity subject to perpetual fluctuations, it follows 

 as a consequence that the variations of the direction must be 



