592 Mr. A. M. Tyndall on the 



occurred in air from the surface of a fine wire, his value of 

 7 mms. instead of 3-4 mras. being attributed at the time to 

 the fact that the average field within 7 mms. of a straight 

 wire was greater than the average field within 3 mms. of a 

 fine point. 



The possible effects of field and current upon z seemed 

 interesting, and the wind pressure apparatus was set up again 

 by the author, to investigate the question. 



The apparatus was that previously used for the work on 

 hydrogen, the experiments now, however, being confined to 

 drv dust-free air. 



Values of jo and z were obtained for various values of current 

 densities C. If it be assumed that in a distance dz ions of 

 one sign only are present, then the specific velocity v of the 

 ions may be calculated from the expression 



I dz 



The typical result is given in Curves 5 for a current of 

 122 microamperes, with the values of p as ordinates and 

 those of z as abscissae. 



It will be seen that the curve is made up of three parts, 

 of which the centre part AB is straight ; it is from this part 

 AB that the values of v in past work have been calculated. 



Omitting for the present the ends OA and BC from the 

 discussion, it is clear that if the above expression holds 

 the slope of AB should be proportional to C. The author has 

 found, however, that this is not the case. Curves 6 show 



the relation between ■— - and C for positive and negative 



discharge, C being measured in microamperes per sq. centi- 

 metre. 



The full curves are the experimental curves and the dotted 

 lines are calculated from the above expressions, assuming 

 that v has constant values 1*32 and 1'84 cms. per sec. per 

 volt cm. for positive and negative discharge respectively. 



It will be seen that as the current increases — falls short 



dz 



of the value necessary for proportionality with C; conse- 

 quently the calculated values of the velocities of the ions 

 increase. 



The readings were all very irregular, and the points given 

 on the curves are actually means of a number of readings 

 differing from one another sometimes by nearly 20 per cent. 



