310 The Rev. T. R. Robinson's Experimental Researches on the 



magnet retains when, after being excited, the current is withdrawn, and it is left 

 with the keeper down, and which I consider to be the phenomenon that promises 

 the most direct information as to the law of the coercive force. This state 

 seems to continue for an indefinite time ; at least I have never found any dimi- 

 nution of its intensity after many weeks. Accordingly, in observing it, I have 

 either left it 10" or during the night. It corresponds to the state 31 — D = C. 

 Dui'ing the previous excitation these forces were of much greater amount, and the 

 force C had aided 31— D against i:^: when the latter is withdrawn, J/ re-unites a 

 portion of the opposite polarities, and decreases in consequence, so of course does 

 D. As to C, there is some reason for believing that it may in the first instance 

 aid this re-union ; and that a certain decrease of 31 is required to develop the 

 molecular action on which it depends in the opposite direction. If so, it will 

 depend not on the Jf which co-exists with it, but a previous one. However, 

 we know that it must also soon begin to oppose the force 31— D. Now it is 

 obvious that if C were constant, the final value of 31 — D, and therefore of A, 

 must also be so ; if it were proportional to 3f, A would vanish ; and if it were 

 in any inverse ratio of it, the difierences of L and A would lessen as they in- 

 creased. On inspecting the Table it does appear that all above No. 15 may be 

 considered of the same value, in its mean 130'68: which amounts to this, that 

 the force C cannot arrest the decrease of the magnetism as long as it exceeds half 

 the maximum A. Is it constant above this, where, as has been shown, it yields 

 equally to excitation in either direction, or does it merely suffer there some 

 abrupt change of magnitude ? For lower values of L it decreases, bearing 

 always an increasing ratio to it ; thus in No. 21 it is nearly half, in No. 23 two- 

 thirds ; and in the negative values this continues to hold, although, as in the 

 case of L, they are long less than the positive. If, while the magnet be in this 

 condition, we pass through its helices a current that would in the ordinary mode 

 give it a force equal to its A, its entire effect is not superadded to the other. 

 Thus I found that, having passed i^= 0'9864, which on this occasion gave 

 A = 125-18, if I passed then F = 0-1395, which would have produced 

 L = 124-46, I had Z'-f A = 169-81 ; so that it only added 44-63. This was to 

 be expected from the principles already explained ; but I cannot so well ex- 

 plain an experiment which shows that a current which can give L — \A pro- 

 duces the same results even if the magnet have residual excitation. If a negative 



