386 Reports on Special Researches 



thus enabling the apparatus to be turned to the wind. The movable brass ring carries a 

 gimbal system which supports the electroscope system, so that the latter remains vertical 

 as the ship inclines. The wire W passes through a hole in a screw E in the amber piece G, 

 and by adjusting the screw, coincidence can be secured between the point of intersection 

 of the gimbal axes, and the point at which the wire bends as the ship inclines. It was ascer- 

 tained experimentally that the change in configuration resulting from tilt was not sufficient 

 to alter appreciably the electrical capacity of the system. 



The air, after being drawn through the space between the cylinders, passes in the direc- 

 tion of the arrows P down mto the observatory, whence it escapes from the windows at all 

 parts of the room. Thus, air which has lost its conductivity in passing between the con- 

 centric cylinders is prevented from entering the apparatus again. The funneled opening H 

 can be readily removed when observations are completed, and the apparatus may then be 

 covered with the box seen in Plate 22, Figure 1, on the roof of the observatory to the right of 

 the figure. During leakage tests, the entrance to the concentric cylinders is closed by a 

 wooden disk and the exit by a shutter 0. 



It will be seen that the main source of leakage in the apparatus is across the amber 

 supports S and G, Figure 22. Quite recently these supports have been replaced by two 

 insulators, each of which is divided into two parts by a guard-ring R maintained at the 

 potential to which the electroscope is charged at the beginning of the observations. Thus 

 leakage occurs only as a result of departure from that potential, and hence is very small. 

 The electroscope itself is not provided with a guard-ring, but its insulation can be protected 

 more thoroughly than that of the amber supports S and G. 



The method of recording the observations and of calculating the results will be under- 

 stood by a reference to the example on page 398. One determination of (say) X+ is made 

 as indicated in the example, and comprises two observations of t. This is then followed by 

 a determination of X_, comprising 4 observations, and finally by another determination of 

 X+ comprising 2 observations.^ The mean of the two values of X+ is then taken as the value 

 appropriate to the mean time of the whole experiment. Days on which there were 

 2 determinations of X_ and 1 of X+ alternated with those on which there were 2 determi- 

 nations of X+ and 1 of X_. The conductivity observations were carried on simultaneously 

 with those of the ionic numbers in a manner which will be clear when the determination 

 of the latter element has been described. 



IONIC CONTENT. 



The usual method of measuring the ionic content of the atmosphere is that due to 

 Ebert. It will be remembered that in this method a stream of air is drawn by a fan through 

 a cylindrical condenser, the inner cylinder of which is connected to an electroscope charged 

 (say positively) to about 200 volts. If W is the volume of an- flowing through the cylinder 

 during the experiment, 5 7 the fall in potential m that time, e the ionic charge, n_ the number 

 of negative ions per c. c, and C the capacity of the whole instrument, then if the potential 

 of the inner cylinder is sufficient to result in all the negative ions being caught, we have 



n_eW = C8V 



One of the chief defects of the method is the fact that the time necessary to obtain a 

 measurable alteration 57 in the electroscope is rather long, amounting, according to some 

 authorities, to as much as 30 or 40 minutes.^ This not only accentuates errors due to 

 leakage, but it introduces uncertainties owing to the variation of the ionic density during 

 the time of the experiment. The reason for the slow ness of the method of course Ues in the 



'Since the method of recording the observations is the same for each set, only one set is shown in the example on page 398 . 

 ^See G. C. Simpson and C. S. Wright, Proc. R. Soc. A, vol. 85, p. 188 1911. 



