OF ARTS AND SCIENCES. 



219 



spark was increased three or four times ; but iu no one set of observa- 

 tions was the spark less than doubled in length. These experiments 

 were rei)eated several times on difleroiit days. 



I found great dilliculty in breaking the primary circuit uniformly, 

 and it was only after a long time that my results became regular. 



I tried breaking from the surface of mercury ; but, although the 

 same relative effects were attained, the actual ert'ect was far inferior 

 to that obtained when two bright copper wires were separated in 

 the air. 



Rowland has shown that a very powerful electro-magnet is not 

 needed, in order to obtain good results in studying Diamagnetism. la 

 order to see whether a small magnet would not do to increase the 

 length of the spark, a small electro-magnet, capable of supporting per- 

 haps two kilogrammes, was set up, and a series of observations taken 

 when the circuit was broken between its poles. In this case, the 

 length of the spark was just doubled. Might not a rather small horse- 

 shoe electro-magnet be advantageously placed upon the Ruhmkoitf 

 coil, so that the primary circuit should be interrupted between its 

 poles ^ 



I next put the electro-magnet in one circuit, and broke another cir- 

 cuit containing the primary of the induction coil between the poles 

 of the magnet. With this arrangement, two sets of observations were 

 taken under slightly different circumstances. The results are given 

 below in the actual space over which it was found that the sparks 

 would just pass. The words " outside " and " inside " denote that the 

 primary circuit was broken outside of the poles and between them 

 respectively. 



Of course the relative increase in the length of the sparks depended 

 upon the relative strengths of the two currents. When the electro- 

 magnet is very weak, and the outside current very strong, subsequent 

 experiments showed that there was little, if any, increase in the length 

 of the spark. 



As a direct means of showing that the extra effect obtained by 

 breaking between the poles is not mechanical, the following method 



