964 Mr. S. Smith on Initial 



positive charges which arrive at C, 0, and D are therefore 

 proportional to the number of molecules ionized by negative 

 ions in the spaces S l5 S 2 , and S 3 respectively. It is clear 

 that with an apparatus designed on these principles one 

 could detect the first indications of the production of ions 



bv collisions for small values of - . 



P 

 The ratio, /jl, of the number of negative ions produced by 

 collisions in S 2 to the total number produced by collisions 



can be found theoretically for the different values of — . It 



V 

 will be seen that for the larger values of a (a being the 



number of negative ions produced by collision by a negative 

 ion whilst moving through a distance of one centimetre in 

 the direction of the electric field) the ratio //, is a function 

 of a, but for the smaller values of a, /j, is independent of a. 

 The latter is the more important case, for by substituting 

 the theoretical value of fi in expressions found from experi- 

 mental measurements values of a can be deduced. 



In the experiments the stream of negative ions was 

 obtained by allowing ultra-violet light to fall on the elec- 

 trode 0, and as it was of great importance that the ions 

 should be given off uniformly over the electrode, a prelimi- 

 nary experiment was made on the silver plate used in the 

 apparatus. The plate and a sheet of copper gauze were 

 used as parallel electrodes, and a narrow beam of ultra- 

 violet light was allowed to fall on different parts of the plate. 

 The saturation currents between the electrodes were found 

 to be very approximately the same in every case, showing 

 that the photoelectric effect was uniform over the plate. 



The top electrodes 1, 2, and 3 (fig. 2) were cut from this 

 plate, which was 4*5 cm. in diameter. The width of elec- 

 trode 2 is 4 mm., and electrodes 1 and 3 are separated from 

 electrode 2 by gaps of -J mm. each. The bottom electrode, B, 

 which is situated 2 cm. from electrodes 1, 2, and 3 and 

 parallel to them, is a silvered quartz plate on which is 

 ruled a grating of length 15 mm. and width 3 mm., the 

 lines of the grating being equally spaced at distances of 

 ■J- mm. B fits into a brass ring, which is supported by brass 

 pillars fixed to the base F. When B is in position the 

 grating is parallel to electrode 2. Gr is a brass guard-ring. 

 The pillars P, P, the block A, and the plugs G and D are all 

 of ebonite. In the brass base is a rectangular gap, S, of 

 length 1*5 cm. and width 3 mm. parallel to electrode 2, and 

 the apparatus was so made that the line of centres of elec- 

 trode 2, of the grating, and of the gap is normal to the 



