F. Dellmann on Atmospheric Electricity. 407 



the same charge from the sphere, so that the deflections differed 

 greatly from each other and always diminished. In the last 

 series, which extended over 3 hrs. 25 mins., the first deflection 

 amounted to 7e>l°, and the last to 13^°. The Table II., accord- 

 ing to which the calculations were made, is not the Table II. of 

 Kohlrausch, but one calculated by me, of which more hereafter. 



The last series already shows pretty clearly the proportionality 

 between the electric loss and the density. On the following 

 morning, however, the effect of the drying had again disappeared; 

 for a series of measurements then gave the following numbers : 



1-03 (5), 1-27 (5), 1-38 (5), 0-97 (5), 1-32 (5) 



1-04 (5), 0-98 (5), Ml (5), 074 (23), 079 (20). 

 Although the smallness of the last two numbers is partly due to 

 the greater intervals to which they correspond, yet on the whole 

 the cause of the irregularity must be sought elsewhere. 



After fruitless endeavours to produce regularity by careful 

 attention to the sheU-lac ring, it was at last withdrawn and re- 

 placed by thin leaves of dried caoutchouc. The measurements 

 now gave the following numbers : — 



0-87 (10), 077 (14), 0-80 (21), 0-82 (29), 0-98 (24). 

 Here is tolerable regularity, so that caoutchouc insulates almost 

 as well as dried shell-lac. Thick patches of moulded vulcanized 

 caoutchouc, to whose surface several fibres still adhered, insu- 

 lated much worse. 



Thin and previously dried leaves of gutta percha were next 

 tried, and found not to be so serviceable as thin dried leaves of 

 caoutchouc; they insulate worse, and ensure less regularity. In 

 fact throughout the whole investigation it was manifest that 

 regularity in the loss of electricity always accompanies good 

 insulation. 



Caoutchouc and gutta percha, used in thin plates, have the dis- 

 advantage, however, of becoming partial conductors in a shorter 

 time after being dried than shell-lac does, and this clearly because 

 the former expose a greater surface to atmospheric vapours ; 

 nevertheless a string of gutta percha appeared to be a worse 

 conductor than good shell-lac. 



In order to reduce the influence of the shell-lac to a minimum, 

 the sphere was suspended by a thin wii-e insulated by means of 

 the shell-lac taken from the destroyed ring ; the effect, however, 

 was not satisfactory. No corresponding regularity was produced 

 even when the transferrer, in its old form, was also taken away, 

 and the charging of the measm-er accomplished by a wire in 

 contact at one end with the sphere, and at the other with 

 the conducting wire of the measuring instrument. There being 

 no doubt as to the ertcct of the slicll-lac in this case, the 



