IIXPEKIMENTS WITH IONIZED All!. 



73 



P; C'is ail insulutiiig glass tube '<SKi ocutims. lung, tlii'ougli which the charge is con- 

 veyed by the wire uh^ to be dissipated iu the condenser. The tube (7 is grasped by 

 the hard-rubber sheath F. 11 is thus in contact with the electrometer, and the 

 capacity of the latter (about 90 cm.) is always lai'ge as compared with the con- 

 denser (negligible here and less than 1 cm. in the experiments below). D is a 

 perfoi'ated copper plate closing the condenser and ]>utting it to earth at K. 



3. [iitrodaclor;! observations. — The early experimtiuts made with this a[i[iaratus 

 (Resonatoi', K6, diameter 27^ = 8.6) ai'e given in figure 3, and they seemed to indi- 

 cate a linear relation of potential and time at least within the first 7 minutes. 

 This constant current was not appreciably modified by stoi>pingthe lower tulndure 

 of the resonator with cotton (removed stoppei', cui-ve 5), nor after vigorously airing 

 it out (2, 4) ; neither is tlieie any marked change of current even after 30 niiuutes 

 (2, 3). The figure also shows the degree of insulation (" Ins.") attained. The 

 currents (rates) expressed in scale pai'ts and seconds would here be .01. 



This interesting result was not, however, borne out by longer i)eriods of 

 observation (20 minutes), as shown in figure 4. Mai'ked cuivature is now in 

 evidence wdiich does not seem I'eferable to conduction leakage. Moi'eover the conduc- 

 tion decreases in succeeding curves, apparently indicating a diminution of the ion- 

 ization due to the phosphorus. The initial rates or cuiTents are, roughly, .015, .013, 

 .012, respectively, for the same condenser. Possibly the activity of phosphorus 

 may be quenched l)y its own emanation in the lapse of time, but the discre[)ancy is 

 liable to be thermal. 



A systematic series of experiments was now carried thiough with receivers 

 ramdnir in diameter from 2R—'?> to 2it = 23 centims., as detailed in table 1 and by 



*"* •••11' 



the charts figures 5 and 6. In all cases of figui'e 5, time {t) m minutes is the abscissa, 

 and the corresponding potential in scale parts {s) of the electrometer, (roughly, 8.4 

 sc. pts. are 40 volts), is the ordinate. The mean rate Ss/St is computed in the table 

 for the first 5 minutes of observation, using leadings made about one minute apart. 

 This curve is drawn in fligure 6 where the abscissas ai-e diameters, 1R, in centims. 

 The same chart contains in its ui)per [lart the dependence of the quantity 

 d (log V)/I{dt on diametei', to be discussed below. In figure 5 the insulation 

 (" Ins.") of the apparatus is .shown. 



TABLE I.— ELECTRIC LEAKAGE OF SPHERICAL CONDENSERS. 



