272 Mr. H. A. Rowland on Magnetic Distribution. 



taken at one point and Q and Q s at another point whose dis- 

 tance from the first is /, we shall have the four equations 



Q =C, CH =-£ ri {ke« + e-), 



Q! Qj 



Calling p- e = H and -~- = G, we shall find, on eliminating 



C and A and developing e rl and e~ rl , 



% 2/GH + l \ 



or, to a greater degree of approximation, 



^=Ka/ i3 ( 3 Sw +1 )- 6 )- • • < 9 *> 



Before applying these formulae to any series of observations, 

 the latter should be freed from most of the irregularities due to 

 accidental causes. For this purpose the following Tables have 

 been plotted and a regular curve drawn to represent as nearly 

 as possible the observations ; in other cases a column of differ- 

 ences was formed and plotted. In either case the ordinates of 

 the curves were accepted as the true quantities. But, for fear 

 that some might accuse me of tampering with my observations, 

 I have in all cases added these as they were obtained. 



The correction is necessary, because small irregularities in 

 the observations will produce immense changes in r 2 . 



Table IV. contains some of the best observations I have 

 obtained. It is from a bar 57 inches long with a helix 1^ 

 inch long in the centre to magnetize it. Each quantity is 

 the mean of six observations, these being made on both ends 

 of the bar and with the current in opposite directions. 



In this Table a source of error was guarded against which 

 I have not seen mentioned elsewhere. When a bar of iron 

 is magnetized at any part and the distribution over the rest 

 quickly measured, on being then allowed to stand some time 

 and the distribution again taken, it will have changed some- 

 what, the magnetism having, as it were, crept down the bar 

 further. Hence in this Table time was allowed for the bar to 

 reach its permanent state. 



