Electromagnetic to the Electrostatic Unit. 303 



nearly constant at the new value. Inasmuch as glass and 

 shellac are poor conductors, the temperature of the graphite 

 resistances cannot safely be assumed to be the same as that of 

 the air within the box, unless the latter has been kept constant 

 for some time. In order, therefore, to avoid all uncertainty as 

 to their values these resistances were determined anew when- 

 ever used, and, if their temperature changed materially, both 

 just before and just after using. They were compared with the 

 resistances of boxes A and B, two arms of a Wheatstone bridge, 

 with a ratio of 99*89, being taken from A. Here is a specimen 

 observation and calculation. 



C Graphite 19°-8 

 May 22. Bridge reading 24,430. Temperatures 1 A = 20°-3 



( B = 20°-6 



_ j 20000 from B = 20,012 at 20° = 20,015 at 20°'6 

 24, U — ( 4430 " A 4443 " 4444 " 20 '3 



24,459 X 99-89 = 2,443,200 ohms at 19 0, 8 = temp, at which used. 

 This value is reliable to within one part in five thousand. 



It is proper to add that if these graphite resistances are put 

 into a circuit where there is a large difference of potential be- 

 tween their terminals, their resistance is immediately dimin- 

 ished by heating. With three Bunsen cells used in measuring 

 their resistance no heating was perceptible. In the determina- 

 tion of capacity there was a difference of potential between the 

 terminals always less than two volts and usually less than one. 

 When the temperatures were maintained constant the resis- 

 tance after use was always precisely the same as before. 

 While, therefore, the use of high graphite resistance is some- 

 what restricted where great accuracy is desired, they still may 

 serve a very useful purpose in many cases, and are the most 

 convenient and reliable of any high resistance, aside from 

 metal wires, that I know anything about. 



Arrangement of the apparatus. 



The vibrators were fixed as near as possible to the condenser 

 to reduce the capacity of the charging wires to a minimum. 

 The condenser, galvanometer and other parts of the apparatus 

 were insulated with great care ; and yet in spite of all precau- 

 tions leakage made its appearance on rainy days, and a slight 

 trace of leakage could usually be detected. Observations were 

 consequently confined to fair weather. The apparatus for de- 

 termining the frequency of the forks was kept always ready 

 for use. 



Sources of Error. 



1. Resistances. — The constant errors in the resistances must 

 have been very small, and corrections for temperature fluctua- 

 tions were made with great care. 



