IN AN ELECTRIC FIKLD — HENDERSON. 125 



expected that in an atinosplicrc of pure nitrogen, say, all the 

 rest-atoms would he similarly eharged. l^xix'rimcnts carried 

 out in atmospheres of pure nitrogen and oxygen gave a|)proxi- 

 mately the same fractions positively charged as for air, 

 wiiieii disj)osed of this suggestion. 



A possible explanation of the results ol)tained by Welliseh 

 might be found in the type of vessel used. In these investiga- 

 tions Welliseh, as well as the others, used as a testing vessel 

 a cylindrical case with an insulated rod fixed centrally inside 

 it, case and rod serving as electrodes. The great disadvantage 

 of this type of vessel is its non-uniform field. A simple 

 calculation will show that most of the drop in potential 

 between the electrodes takes place near the rod. The 

 presence of ionization still further weakens the potential- 

 gradient in the bod}^ of the vessel and increases it near the 

 rod. In the corners at the ends of the cylinder and near 

 the insulating plug, the field is much weaker than even in 

 the body of the vessel. Now, although the average value 

 of the field within the vessel maj^ be such as apparently to 

 prevent recombination, there is a part of the volume of 

 the vessel in which the field is very much weaker than the 

 average and in which recombination might take place to 

 a considerable extent. 



To avoid this and other difficulties, the vessel used 

 in the present experiments consisted essentially of two 

 parallel plates surrounded by guard-rings. This type of 

 vessel gives a much more uniform potential-gradient than 

 the cylindrical type. The guard-rings prevent distortion 

 of the field at the edges of the electrodes and also allow 

 the use of insulating material on their outer edges without 

 effect on the field between the real electrodes. 



XL- Apparatus and Method. 

 The testing-vessel used in these experiments was a 

 slight modification of the "second type" of vessel, described 



