Mr. W. Spottiswoode on Electro-magnetic Rotation. 391 



greater ; but with carbonic acid greater steadiness of position was 

 obtained. With both ether and chloroform occasional flashes, bril- 

 liantly illuminated, were seen ; but some chemical action appeared 

 to take place militating against the steady development of the flame- 

 sheet. With coal-gas there was an inconvenient deposit of carbon 

 upon the sides of the chamber. With hydrogen the cloud was 

 not sufficiently developed. 



The success which attended the experiment with air may pos- 

 sibly be partly due, as suggested above, to the combustion of 

 the extraneous matter floating therein ; and in fact the brilliancy 

 and extent of the sheet may be increased by attaching a piece of 

 metallic sodium to the negative terminal, or by causing a stream of 

 any of the chlorides in powder, e. g. of strontium, lithium, &c, to 

 flow across the field of action. 



When a piece of sodium (or better still of soda) is attached to 

 one of the terminals, two effects may be noticed. When that ter- 

 minal is negative the whole of the flame is bright yellow, showing 

 that the sodium is not only detached but even carried across the 

 field and deposited on the positive terminal. When, however, the 

 terminal to which the sodium is attached is positive, it is f ound that 

 the flame, when observed through a red glass, appears yellow to a 

 certain distance from that terminal, but red beyond, and also that 

 the pitch of the helix is less near the positive than near the nega- 

 tive terminal. These effects may be attributed to the presence of 

 metallic vapour evolved by the heat at the positive terminal, but 

 not carried across the field as. when the terminal in question is 

 negative. 



The following explanation of the phenomenon is due to Prof. 

 Stokes, from whose correspondence it is substantially taken. The 

 mathematical solution, although only roughly approximate, is per- 

 haps still quite sufficient to give the general character of the ex- 

 perimental results. 



The magnetic field will be supposed uniform, and the lines of 

 force parallel straight lines from pole to pole. The path of the 

 current when undisturbed is also a straight line from pole to pole. 

 In such a condition of things, every thing being symmetrical, no 

 rotation would take place. But if through any local circum- 

 stance, as in the experiment in air, or through heating of the 

 chamber as in the exhausted tube, or otherwise, the path of the 

 current be distorted and displaced, then each element will be subject 

 to the action of two forces. To estimate these, let ds be an ele- 

 ment of the path, with rectangular components dec, dy, dz, C 

 the strength of the current, and R the magnetic force with com- 

 ponents X,T, Z, which in the first instance will be treated gene- 

 rally. Then one force will be that tending to impel the current 

 in the direction of the axes respectively, and may be expressed by 



C(Ydz~Zdy) : ds, C(Zclv-Xdz) • ds, C(Xdy-Ydx) : ds. 

 Besides this, there will be the tendency of the current to follow 

 the shortest path so as to diminish the resistance. Representing 



