226 Mr Townsend, Secondary Rontgen Rays. 



^-millimetre thick. The sheet of aluminium between the bulb 

 and the window was removed so that the strength of the primary- 

 rays passing through the gas should be the same in both cases. 

 The results of the observations made with this arrangement are 

 shown in curve II. figure 4. 



It is easy to explain the form of these curves on the sup- 

 position that the rate of absorption of radiation by a gas is 

 proportional to the ionization. When the pressure exceeds a - 

 certain value, the effect of surface rays will be located near the 

 surface which emits them, and for greater pressures the number, 

 of ions produced will be the same, but they will be generated 

 nearer the surface. The effect of the primary rays is proportional 

 to the pressure of the gas, so long as the pressure is not big 

 enough to reduce their intensity appreciably (it would probably 

 require a gas at many atmospheres pressure to produce this effect, 

 so that we will consider that the ionization contributed by the 

 direct rays is proportional to the pressure). Under these circum- 

 stances the number of ions produced in a given volume will be 

 S+pD where 8 is the effect of the surface rays, p the pressure, 

 and D a constant depending on the direct rays. The quantity 8 

 is practically constant for values of p greater than a certain 

 pressure ir, which is about 50 centimetres of mercury in the 

 present case. We conclude that if there were no surface effects, 

 that the curve representing the connection between pressure and 

 the total number of ions produced, would be a straight line D 

 through the origin parallel to the portions of curves I. and II., 

 which correspond to pressures greater than 50 centimetres. 



When the pressure is reduced the rate of absorption of the 

 secondary rays is reduced proportionally, so that when p is less 

 than 7r the surface rays extend across the vessel containing the gas 

 and part of their energy is absorbed by the opposite wall. The 

 number of ions they generate in the gas will be no longer constant, 

 but will diminish with the pressure, so that the formula 8 + pD 

 would no longer hold on the supposition that S is constant. The 

 exact connection between 8, p, and it, would depend on the shape 

 of the vessel containing the gas. 



This general explanation shows that we ought to get curves 

 similar to those in figure 3 by subtracting from the ordinates of 

 the curves in figure 4 lengths equal to the ordinates of the 

 straight line D. For complete agreement it would have been 

 necessary to have made the experiments on the variation with 

 pressure with the apparatus shown in figure 2, but even with the 

 results before us we can see that there would be a fairly good 

 agreement. 



In conclusion I desire to express my thanks to Professor J. J. 

 Thomson for many suggestions in connection with this investigation. 



