Atmospheric-Electric Observations, 1915-16 



385 



Then the total loss of electricity in the time t is Ci5V, and the loss by leakage is Cit5V/9, or 

 if we call t the time which the apparatus would require to leak through the range 5V, the 

 leakage term is CitSV/L Thus from (4) 



Ci8V{T-'-t-')=4Tr\C2V 



(5) 



A determination of t is made just before the first and just after the last of the determina- 

 tions of T con-esponding to each set of observations for the unipolar conductivity, and the 

 mean value r^ of r' is used in the formula. It is easy to see that the mean of the recip- 

 rocals rather than the reciprocal of the mean should be used. In determining the leakage 

 term it is desirable to make observations over a range of potential in the neighborhood of 

 the midpoint of the small range 57, and the value substituted for V in (5) should be the 

 mean value corresponding to this range. 



Fig. 22. — Diagram of Conductivity Apparatus. 



In order that the theory of the apparatus may apply properly, it is necessary that the 

 air-flow shall be sufficiently great, and it is desirable that it shall take place for a longer 

 period than is readily possible with clockwork devices. For this reason the fan is run by a 

 small electric motor driven by a 30- volt battery. A diagrammatic view of the conductivity 

 apparatus is shown in Figure 22. The supporting-rod A of the central cylinder rests in a 

 brass socket fixed in the amber plug S, and the amber plug is contained in a brass tube 

 supported from the brass wall BB by three brass struts. The upper part of the apparatus 

 is inclosed by the box CC, the whole of which is above the roof of the observatory. To the 

 bottom of the box is fastened a brass ring which can turn in a fixed ring screwed to the roof, 



