324 Mr. H. A. Rowland on the Magnetic Permeability 



in my first paper, where the cnrve of permeability was continued 

 beyond the point to which the experiments were carried. 

 Neither does experiment up to the present time furnish any clue 

 as to whether it is 23 or 3 which attains a maximum. 



As the matter is in this undecided state, I shall hereafter in 

 most cases calculate both 3 and k as well as 23 and ft, as I am 

 willing to admit that 3 may have a physical significance as well 

 as 53, even on Faraday's theory. 



There is a difficulty in obtaining a good series of experiments 

 on nickel and cobalt which does not exist in the case of iron. 

 It is principally owing to the great change in magnetic permea- 

 bility of these substances by heat, and also to their small per- 

 meability. To obtain sufficient magnetizing-force to trace out 

 the curve of permeability to a reasonable distance, we require at 

 least two layers of wire on the rings, and have to send through 

 that wire a very strong current. In this way great heat is de* 

 veloped ; and on account of there being two layers of wire it 

 cannot escape ; and the ring being thus heated, its permeability 

 is changed. So much is this the case, that when the rings are 

 in the air, and the strongest current circulating, the silk is soon 

 burned off the wire; and to obviate this I have in these experi- 

 ments always immersed the rings in some non-conducting liquid, 

 such as alcohol for low temperatures and melted paraffin for 

 high temperatures, the rings being suspended midway in the 

 liquid to allow free circulation. But I have now reason to sus- 

 pect the efficacy of this arrangement, especially in the case of the 

 paraffin. The experiments described in this paper were made at 

 such odd times as I could command, and the first ones were not 

 thoroughly discussed until the series was almost completed; 

 hence I have not been so careful to guard against this error as 

 I shall be in the future. This can be done in the following 

 manner — namely, by letting the current pass through the ring 

 for only a short time. But there is a difficulty in this method, 

 because if the current is stopped the battery will recruit, and 

 the moment it is joined to the ring a large and rapidly decreas- 

 ing current will pass which it is impossible to measure accu- 

 rately. 1 have, however, devised the following method, which 

 I will apply in future experiments. It is to introduce into the 

 circuit between the tangent-galvanometer and the ring a cur- 

 rent-changer, by which the current can be switched off from the 

 ring into another wire of the same resistance, so that the current 

 from the battery shall always be constant. Just before making 

 an observation the current is turned back into the ring, a read- 

 ing is taken of the tangent-galvanometer by an assistant, and 

 immediately afterward the current is reversed and the reading 

 taken for the induced current ; the tangent-galvanometer is then 



