DowLiNG — Stead;/ and Turhdent Motion in Gases. 379 



amount of " mixing " takes place. The air-particles will frequently be 

 brought near the sides of the tube. We should expect this to aid the 

 processes which destroy the conductivity. 



If we plot a curve showing the variation of residual electrical conductivity 

 at B, with tlie rate of flow of air through tlie tube, we find that the con- 

 ductivity at first increases "smoothly," as we sliould expect, with the 

 velocity— but at a certain velocity there is a well-pronounced discontinuity. 

 After tliis occurs, the curve may continue to show an increase of conductivity 

 with velocity, as before, or, in some cases, a drop, as explained later. 



In what follows, instead o£ the volumes flowing per second through a 

 tube, we shall speak of the mean velocity of the gas. If V denotes this mean 

 velocity, and if v is tlie volume passing per second through a tube of radius a, 

 obviously we have 



F = A . . . (2) 



Trrr 



Also if Vmax is tho maximum velocity (on the axis) for stream-line flow, 



V^ax = 2 F. 



Pari' III. 



Verification of law connecting the Critical Velocity with the Tube Diameter. 



Tlie first experiment, made principally with a view to verifying the 

 suitability of the method, was carried out as follows : — 



The tube AB (of glass, 75 cms. long, and 1'6 cms. internal diameter) was 

 connected througli BC, as described, to a gas meter, and an electrically 

 driven fan was utilized to draw air through the system ABC and the meter. 

 The end of the tube AB was situated just at the point A, and the air was 

 ionised as it entered tlie open end. Variations in the velocity of the air 

 were obtained by altering the speed of the fan-motor ; and the rates of flow 

 were calculated from the quantity of air registered by the gas meter in a 

 minute. The mean velocities of the air, plotted against the ionisation 

 measured, are shown in curve 1, Plate XXX IL, the electrometer readings 

 being taken with the capacity K = '01 microfarad. 



A decided discontinuity occurs at a mean velocity of 232 cms. per second 

 (corresponding to '97 cubic feet per minute). The break is extremely well 

 marked, and was, bej^ond doubt, not in any way due to the action of the 

 gas meter, which was well within its range of working. The mean diameter 

 being V6 cms., the product 



r„ X (/=232x 1-6 = 371 . . ., (C) 



and is in good agreement with the values {A ; B) already referred to. 



This agreement may be taken as an imlirect proof of the validity 



