Mass of the Ions in the Electric Wind in Air. 409 



larities, and I think that curves III. afford, on the whole, 

 considerable support to the premises of equations iv. and v. 



It remains to deduce V, the velocity of the ions in unit 

 electrostatic field. 



For + discharge cfc/^P = 0-0474 centim. per dyne. 

 „ - „ „ =0-0687 „ „ 



C = 2'37microamperes = 7100 E.S. units. 



Hence 



V+ =7100x0-0474 = 336 centim. per second. 

 V- = 7100x0-0687 = 489 „ „ 



The sum of these values is 825. Rutherford's value in the 

 same units was 960. Their ratio is 1*45. Zeleny's value 

 was l - 25. 



Experiments by the "Point and Ring' 3 Method, 



I was at first inclined to attribute the above differences to 

 the fact that the ions studied hy Rutherford and Zeleny had 

 moved in very much weaker fields than those I used ; but 

 there is another way of accounting for them. The symbol 

 P stands for the drag of the ions on the whole gas between 

 point and plate. If this is all received by the plate as fluid 

 pressure the gauge measures P ; but if there is anything like 

 a rebound of the wind from the plate, the pressure on the plate 

 may or may not be too great according as the momentum 

 of the rebound is received on a fixed body or is allowed to 

 push back the atmosphere above the plate. Now nearly all 

 the readings were taken with the disk D in position, and it 

 seemed possible that D ha 1 played the part of this fixed body 

 and had thus given rise to an error of excess in the readings 

 of the gauges. Such excess would vanish for very small as 

 well as for very large values of z ; the pressure-curve would 

 therefore still cut the pressure-axis at the proper point, but 

 its slope would be too great for values of s which were small 

 compared with the diameter of D ; and such having been the 

 case for curves III., the values of V obtained from these 

 curves might be expected to be too small. 



By arranging that the gauge measured the difference of 

 pressure between two opposite points on D and on the plate, 

 the error could of course be avoided ; but considering the 

 uncertainties attending the pressure integration and the fact 

 that each separate point in the curves represented several 

 hours of work, it seemed desirable, if the curves were to be 

 repeated, to improve the method. Some plan was obviously 

 needed whereby all this labour might be replaced by a single 



