NATURAL PHILOSOPHY. 119 



in the circuit some water holding a little salt in solution, and in 

 which two plates of metal were immersed. As soon as the principal 

 current was weakened or stopped the inverse current was exhibited 

 by the plates. 



" In order to reproduce all the details of the natural phenomena, I 

 caused an apparatus to be constructed composed of a sphere of wood 

 about ten inches in diameter, which represented the earth, and car- 

 ried at each pole a bar of soft iron about two inches long and about 

 one inch in diameter. Each bar rested on a vertical cylinder of soft 

 iron to which it was united, and thus the sphere was supported. So 

 arranged, the sphere had a horizontal axis terminating in two appen- 

 dages of soft iron which could be magnetized by bringing the two 

 cylinders on which they rested in contact with the poles of an electro- 

 magnet, or by surrounding the cylinders with coils of wire traversed 

 by electric currents. Each of the iron bars was surrounded by a 

 glass cylinder (mancJion) between five and six inches in diameter, and 

 about seven inches long, and in which it occupied an axial position, 

 projecting into the middle of the glass. The two vessels were her- 

 metically sealed by two metallic caps, one of which was traversed by 

 the iron bar, while the other carried a metal ring upon two arms, the 

 centre of the ring coinciding with the end of the iron bar, and hav- 

 ing its plane perpendicular to the axis of one bar. The diameter of 

 the ring is a little less than that of the glass. Stopcocks were con- 

 veniently placed to allow of a vacuum being formed in the glass vessels, 

 and any kind of gas introduced. 



u To use this apparatus, the wooden ball is covered with two strong 

 bands of bibulous paper, one occupying its equator and the other 

 crossing it from pole to pole, and making contact with the two bars 

 of iron. On this last band, pieces of copper about one or two thirds 

 of an inch square are fixed at equal intervals with copper tacks that 

 penetrate the wood. All the copper squares are arranged in the 

 same meridian. Between two of the squares a metallic communica- 

 tion is established with the thread of a galvanometer placed about 

 twelve yards off, so that its needle shall not be directly influenced by 

 the electro-magnet. Having thus arranged the apparatus, the paper 

 bands are moistened with salt and water, and the equatorial band is 

 connected with the negative electrode of a Ruhmkorff's coil, which 

 has its positive electrode brought into communication, by means of 

 a bifurcated wire, with the two metallic rings which are inside the 

 glass vessels, and in highly rarefied air. The discharge is soon seen 

 as a luminous jet between the rings and the extremity of the iron 

 bar, sometimes in one vessel, sometimes in the other, but rarely in 

 both at once, although both are placed under apparently the same 

 circumstances. 



" As soon as the soft iron is magnetized, the jet spreads and forms 

 an arc round the central wire, animated by a rotary movement, the 

 direction of which depends on the character of the magnetization. 

 It is evident that it depends also on the direction of the discharge, 

 but we have supposed this direction constant, and resembling that of 

 nature, that is to say. directed from the circumference towards the 

 centre. It is important to notice that if the air be not too rarefied, 

 a moment is observed in which, when the iron bars are magnetized, 



