428 Dr. A. M. Tyndall on the Spread of the 



While maintaining "constant spread" it was not possible 

 to vary the current more than about 100 per cent. Within 

 these limits, however, r was independent of current ; the 

 above values are mean values. 



The mean value which, omitting the first reading, is 1*39, 

 is in agreement with that obtained by recognized methods^ 

 namely about 1*37, when, as in the above work^ no refined 

 process of drying of the air has been employed. 



Now there are two assumptions made in the above treat- 

 ment : (a) that there is no back discharge from the plate, 

 and (b) that the starting conditions close to the point are the 

 same for both signs of discharge. 



It is noi possible to calculate the effect of back discharge 

 on the distribution, but it is probable that the presence of a 

 small amount of back discbarge in both signs of discharge 

 would have little effect upon the value of r obtained, 'the 

 rise of r at short distance (see Table I.) might be taken as 

 evidence of a strong back discharge under these conditions 

 were it not for the probability that at these distances the 

 assumption (b) breaks down. 



The different appearance of the glow at a positive point to 

 that at a negative, the modified nature of the discharge within 

 a few millimetres of the point, as shown by the electric wind 

 measurements and the recent experiments of Moore, all lead 

 to the conclusion that the conditions close to the point are 

 different from those further awav. These may well differ in 

 the two signs of discharge. For short distances between 

 point and plate this difference is sufficient to introduce an 

 inaccuracy in the ratio r, though it appears negligible for 

 the greater distances. 



The results may therefore be explained without assuming 

 unv hack discharge io be present. On the other hand, the 

 fact that r is correct in cases where the discharge region 

 was faintly luminous makes it possible that it is present 

 in small quantities in both signs of discharge without 

 affecting the relative distribution of current in the two 

 cases. 



The breakdown of one or other of the assumptions (a) and 

 (b) makes it impossible to apply the method to different 

 pressures. Thus the pressure was varied from atmospheric 

 down to 10 cm. : within these limits the product of the 

 mobility (v) and pressure (p) should be constant. Headings 

 were taken of constant spread at various pressures, one of 

 which was atmospheric, and the values of vp obtained in terms 



