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



concerned. They all comprise three terms affected with ^, and, of course, show a 

 decrease at first to a minimum, then an increase to a maximum, and a subsequent 

 decrease. Now, as I formerly noticed, the L represents the polarity at the 

 contact surfaces of the magnet and keeper ; this depends on, first, the polari- 

 zation of those molecules which are excited by the helices; secondly, on that of 

 the remainder of the magnetic circuit, and therefore its amount depends on the 

 intensity of these polarizations, and on the facilities with which their influence 

 is transmitted by induction. The intensity, again, depends on the suscepti- 

 bility of the molecules directly, and inversely as the coercing force. Now each 

 of these maybe expected to change with the temperature. The correlation 

 of heat with other molecular forces is such that, a priori, we would anticipate 

 its lessening such forces as we are considering ; and we find that at a red heat 

 even iron is scarcely attracted by the most powerful magnets, which must de- 

 pend on the molecules ceasing to be excitable. A diminution of this power, 

 and of that which transmits the magnetism from one particle to another, must 

 lessen L, while a contrary effect will arise from the diminution of the coercing 

 force. All these influences will be functions of the lengths of the magnetic 

 current, and of the excited cylinders ; and accordingly we find that the co- 

 efficients in (12), (13), and (14), increase with the latter. Calling them a, fi, 

 and 7, and the length of the cylinder z, the three values of 7 are exactly repre- 

 sented by the formula a + bz, and those of a nearly by a' + b'' ; from which one 

 might infer that 7 corresponds to that part of the change which belongs to the 

 excitabiles, and a to the conduction. I have not found any simple expression 

 for /3, but since it gives the intermediate increase of magnetic attraction, which 

 (as I hope to show in a future communication) depends on the coercive force, 

 we may refer it to that. 



I did not think it necessary to investigate these corrections for the other 

 lengths of cylinder, as these three give sufficient data for interpolation, and 

 from them,* 



• Although it is not safe to interpolate beyond the limits of the observations, yet, computing 

 from these the constants for twelve-inch cylinders, and reducing by them the temperature ex- 

 periments with my first magnet, I find the higher gives in each pair an L' less than that of the 

 lower by 7'56, 6'92, 4-44, 6-24, or in the mean 6'29 for a difference of temperature = 101° 6. This, 

 therefore, shows both that the same law holds in this magnet, though hollow, and that these con- 

 stants will serve to reduce the observations made with it. 



