200 Dr. E. D. Kleeman on h Mays. 



opposite direction to its initial motion, its chance of recom- 

 bining with its parent molecule is greater than when the 

 field is in the opposite direction* This would give rise to an 

 asymmetry in the collision leaks in the same direction as the 

 effect just considered* 



A simple consideration will show that the asymmetry should 

 increase with the distance of separation of the plate and foil. 

 The collision leak due to the ionization of a slab of gas of 

 thickness dec midway between the electrodes is equal to 



- a 



Q>e' A when the field is in one direction, and roughly 



Ce^ 2 ' when the field is reversed, using the usual 

 notation. The difference between the leaks thus rapidly 

 increases with x, the distance between the electrodes. It 

 would also depend on the nature and pressure of the gas. 

 The asymmetry may thus be very large under favourable 

 conditions. I found, for example, in one of my experiments 

 when the chamber contained hydrogen at a pressure of '8 cm. 

 of mercury, that the leak changed from 928 to 1592 on 

 reversing the field, the difference of potential between the 

 electrodes, which were separated by '5 cm., being equal to 

 490 volts* Dr. Campbell has informed me, in a private 

 letter, that he also gets a great change in the magnitude of 

 the leaks on reversing the field when it is very large. He 

 finds it especially noticeable when the field is so large that 

 the positive ions produce ions by collisions This fits in with 

 what we would expect from the foregoing considerations. 

 Any effect that increases the magnitude of the collision leak 

 increases the asymmetry. Kow the leak is greatly increased 

 when ionization by collision of the positive ions begins to 

 come in s It is equivalent to supposing that a sudden in- 

 crease in the value of a takes place, winch, as shown by the 

 expression given, would produce a great increase in the 

 difference between the leaks. 



The question now arises why there appears to be no 

 appreciable difference in the number of electrons given off 

 from the incident ;ilid emergent surfaces of the electrodes. 

 If the component velocity of the electron in the saine direc- 

 tion as the motion of the a particle is small, the effect may 

 easily be masked by the various disturbing effects which 

 exist* Thus my experiments and Dr. Campbell's suffer in 

 this connexion from two defects, which are, however, un- 

 avoidable. In order to use as large a number of a particles 

 as possible a beam is used in which the a particles are not 

 parallel to one another* Now it can be shown than when an 



