DowLiNG — Steady and Turhdenf IfoHon in Gases. 



397 



opening in the base directly beneath the hook on the end of the beam. 

 A glass tube B was fitted with the wire frame C, and thereby suspended 

 from the balance by a fine wire. A bent pipe P of almost the same diameter 

 as B was held in the position shown, and 

 accurately adjusted so that the paper tube, 

 projecting from inside it, passed about a 

 centimetre down inside B without toiiching. 

 Thus the tube B was free to move with tlie 

 swing of the balance-beam. The pipe P 

 was connected to a gasometer of the pattern 

 already described, from which a measured 

 blast of air could readily be obtained by 

 adjusting the stopcock and allowing the 

 drum to fall by its own weight. The ex- 

 periment consisted then in counterbalancing 

 the downward pull on B by weights on the 

 sealepan. No allowance need be made for 

 differences in the densities of air inside B, 

 and outside ; for the effect of these will be 

 small in comparison with the quantities we 

 are measuring. 



The weights on being plotted against the 

 mean air-velocities in the tube give a curve 

 at first almost straight and then gradually 

 bending upwards. At one point there is a 

 distinct break in the curve — the rate of 

 increase of resistance becoming suddenly 

 less. This occurs at a point agreeing very 

 well with our expectations; for a 1*9 cm. 

 tube it is at a velocity of about 70 cms./sec, Fig. 4. 



while for a 2-5-cm. tube it occurs at about 84 cms./sec. The expected values 

 are 66 and 81 respectively (cf. curve, fig. 3). The general form of the 

 curve and " break " seems to indicate that a change of some kind has 

 taken place in the nature of the flow. Consequently we can assume 

 that the production of the second " break " of the former experiments may 

 really be due to some cause of the nature indicated previously. 



In conclusion, it is interesting to inquire whether the numerous experi- 

 ments on gaseous ions may not in some cases be influenced by the plienomena 

 we have been discussing. In such experiments the flow of gas is usually 

 constant, whereas we have used different velocities and a constant electro- 



From Gasometer 



