all knoTon Substances to the Magnetic Influencey S^c. 105 



These results, however, though sufficiently uniform and con- 

 sistent for most practical purposes connected with the present 

 investigations, were not so entirely satisfactory as to be received 

 as conclusive. For the difference between the observed and cal- 

 culated deviations at the first distance was by far too considerable 

 to be ascribed to any error of observation or other accidental cir- 

 cumstance, since it was proved to be occasioned by some con- 

 stantly acting cause, because of a similar discrepancy being al- 

 ways found to occur through an extensive series of experiments 

 with many diff'erent magnets. 



The discrepancies beyond the mere errors of observation were 

 at length discovered to be mainly owing to the adoption of ex- 

 act lengths of the magnets as integers of distance, instead of the 

 length of the interval between what may be termed the foci of 

 attraction in the north and south portions of the bar. For, al- 

 though the principal energy of a well constructed and well mag- 

 netized bar is doubtless at the extremities, yet a considerable 

 though rapidly diminishing power necessarily prevails within 

 the extremities. The aggregate or resultant action, therefore, 

 of the varying intensities of either half of the bar may be re- 

 ferred to a particular point or focus, which, through the ap- 

 plication of the foregoing law of attraction, may be determined 

 by approximation experimentally. 



In order to determine the position of the foci of attraction 

 in a three-feet bar-magnet (A) of very superior construction, I 

 assumed, in the first instance, the Jbcal length to be 2 feet 10 

 inches, considering the foci of attraction to he an inch within 

 the extremities of the bar. In this case the '* Diiference'' in the 

 first length, which in the preceding instance was 2° 14', (Col. 9.) 

 was now reduced to less than one-half. A focal length of 2 feet 

 9 inches was then tried, wben a further reduction of error was 

 observed. Lastly, I assumed the distance between the foci to 

 be 2 feet 6 inches, and then the results were most strikingly 

 consistent and satisfactory. 



The following Tables exhibit the results of the first and last 

 •eries of these experiments. 



