Negative Ions hy Collisions at High Pressures. 295 



The ordinates of the rest of the curve then gave n for the 

 corresponding X. 



An inspection of his curves shows that this method is 

 open to grave objection, for in no case is the saturation 

 curve parallel to the axis of X, and it is therefore very diffi- 

 cult to tell where the saturation curve ends and the ionization 

 curve begins. The possible percentage error in n is large, 

 and therefore the possible error in a. is also large. 



In order to test the accuracy of the results thus found in 

 the case of air, some of his values of a have been redeter- 

 mined by Townsend's method, using the actual pressures and 

 electric forces employed by Bishop. 



The currents between two parallel plates were measured 

 in the ordinary way (see ' Theory of Ionization of Gases/ 

 page 5), the distances chosen being 1, 2, 3, and in some 

 cases 4 mm. 



The ratios of successive readings should therefore be con- 

 stant, and if r is the value of this constant, 



a = 10 log e r. 



An investigation of a table of natural logarithms shows 

 that for values of r up to about 1*6 the variation of log e r is 

 extremely rapid compared with the variation of r, and therefore 

 a slight error in the determination of r involves a large 

 error in a. 



For instance, if 



r = 1-1, a = -95, 



and if 



= 1'2. 



1-8, 



No attempt was therefore made to measure the lowest 

 values of a. given by Bishop, while, on the other hand, suf- 

 ficient volts were not obtainable to measure his highest 

 values. 



The results obtained were (for air) : — 



Pressure 

 m cms. 



Force in 

 volts per cm. 



Ratio r. 



Value 

 a. obtained by 

 Bisbop. 





1806 

 193o 



1-92 

 •2-47 



652 



904 



797 



4-16 

 5-21 

 306 



2-64 



492 



3010 22 













The values of the ratio r were in all cases found to agree 



