58 Proceedings of the Royal Irish Academtj. 



uniformly distributed force would maintain the filament of air in question 

 in quite rapid motion can be easily shown. 



XV. — Suppose u is the estimated speed of the air-current in the smoke 

 filament, and i is the electric " transport " current measured by the electro- 

 scope, clearly 



{anu) q =i, 



where a is the cross-section of the smoke filament. 



In an actual case, for example, this equation gave a value {nau) = 5 x 10*. 



Considering a cylinder 1 cm. long of the smoke filament, exposed to a 



V 

 field - = 40 volts per cm., we have, if q = 10s : — 



^ a.n .q.V 5 x 10'^ x 10 x 4-8 x IQ-i" x 40 , 



/ = -, ~ oTTTi dynes 



I u X 300 



= 3x10"' dynes, if m = 1 cm. per sec. 

 This is equivalent in its effect to a pressure difference p between the ends 

 of the cylinder if a^j =/. Comparing the actual case to that of a tube of 

 section a, in which air is kept in motion by a pressure slope p per unit length, 



€C7J 



U = average velocity of air = - — 



OI/TT 



3 X 10-5 . 



= •" cm. per see. 



8 X 180 X 10-" X 3-14 



Thus u and U are of the same order, and it is clear that the force / is 

 sufficient to maintain the whole air-stream in motion with the actual velocity 

 observed.! 



XVI. — Our supposition finds support from the observed behaviour of 

 these smoke filaments. Their sharpness and obvious tendency to resist 

 rupture indicates some internal structure conferring a sort of cohesion ; the 

 tendency of the particles to remain in the moving air-stream as affording a 

 path of least resistance is in keeping with these properties. Again, in the 

 " visual " experiments, where the smoke filament had to traverse a wire 

 gauze screen, it invariably made for the centre of an opening, and never 

 showed a tendency to strike the wires or spread out. When, in making a 

 ■change in the fields, the voltage Fg was removed, the smoke rose in the space 

 £G (fig. 1) vertically. On reapplying the field this vertical thread did not 

 move transversely to a noticeable extent. The conditions in this case would 

 be more in keeping with Millikan's experiment, and the velocity would be of 

 a very small order. 



' The possibility of this process occurring in the atmosphere raises an interesting 

 question, which might be worth consideration. It also suggests a possible source of 

 error in Wilson's original determination of e. 



