96 ME. GEOEGE C. SIMPSON ON THE 



errors of reading were avoided and errors due to the unsteadiness of the leaves 



greatly diminished. 



After having by this method obtained a measurement, using, say, a positive charge, 

 the observation was repeated using a negative charge, and finally another observation 

 with a positive charge. The mean of the two positive values, together with the 

 negative value, were used as the result of the whole observation. This method I 

 found to be absolutely necessary if reliable values of the ratios q and r were to be 

 obtained, for both dissipation and ionization undergo great changes in the course of 

 the time taken to make an observation. A whole observation when taken in this way 

 occupied an hour and a quarter. 



Long experience taught me to know when I could expect difficulties with the 

 insulations. On such days, instead of the method sketched above, an observation 

 was taken with one charge, and after that the insulation tested for 15 minutes, 

 then an observation with the other charge, followed by a final insulation test for 

 the same length of time, the whole observation taking about an hour and a half. 



During the summer I had great difficulty in using the Ebert instrument owing to 

 the mosquitoes being drawn into the instrument and so discharging the electroscope. 

 In June the mosquitoes and other small flies were so numerous that it was quite 

 impossible to use the Ebert instrument without some means of keeping the flies out, 

 so I attached a funnel-shaped net to the front of the aspirator tube and used the 

 instrument so protected. I expected that this net would cause some reduction in 

 the value of the ionization as measured by the instrument, so as soon as the 

 mosquitoes were sufficiently reduced in number to allow of observations being made 

 with the unprotected tube I made a series of observations to find the effect of the 

 net. Much to my regret and disappointment I found that the effect of the net 

 varied very much according to the wind strength. In perfectly still air the net 

 reduced the ionization by nearly a quarter, while with a stiff" breeze it had no effect. 

 This made individual observation practically useless, and in all the above tables 

 connecting ionization and the meteorological elements all the observations taken 

 when the net was in use from June 9 to August 12 have been neglected. 

 As the result of a long investigation I concluded that 10 per cent, added to the 

 results in the bulk would just about correct for the effect of the net. Eesults so 

 corrected are used in the curves and tables showing the yearly course of the 

 ionization. 



Radio-activity. In my measurements of the radio-activity, as stated above, ELSTER 

 and GEITEL'S method was used. In order to charge the wire to a negative potential 

 of between 2000 and 2500 volts, I used a small influence machine, built on the 

 principle of a Kelvin replenisher and driven by a falling weight. By means of a 

 variable high resistance, consisting of a strip of ebonite, one side of which had been 

 rubbed with a black-lead pencil and so mounted in a tube that an earth -connected 

 pad could move along it, the potential of the wire could be very easily regulated. 



