1863.] 687 



instead of hanging freely by a glass fibre, is attached to the middle 

 of a fine wire of platinum, whose two ends are secured in such a 

 manner as to keep it always tight. The needle and repelling plates 

 are put in connexion with the body to be tested, the cage being con- 

 nected with the inner coating of the jar, an arrangement which, 

 though opposite to that adopted in the Station Electrometer, re- 

 quires the same formula to be applied in reducing the readings. 



7. The best possible test of the accuracy of the method of reduc- 

 tion above described is to observe the potential of the same conductor 

 with two electrometers, and try whether their indications (thus 

 reduced) differ in a constant ratio when the charges of their jars are 

 varied. I have applied this test, giving charges of different strength, 

 sometimes of the same kind and sometimes of opposite kinds, the 

 conductor tested being a short wire of copper or brass connecting 

 the electrodes of the two instruments. These experiments were 

 performed for the purpose of ascertaining the ratio in question, with 

 the view of reducing all my observations to a common standard ; and 

 I have also investigated this ratio by connecting the jars of the two 

 instruments, and taking earth-readings; but this method is less 

 convenient because it involves the admission of undried air into the 

 jars, thus impairing insulation, and rendering accurate observation 

 difficult ; whereas in the former method of observing, if performed 

 in favourable weather, the insulation is as good as perfect. As it is 

 important to establish practically the accuracy of the method of re- 

 duction pursued, I give in extenso all the comparisons that I have 

 made by the methods just described. 



8. In using the Station Electrometer, I have given a fresh charge 

 to its Ley den jar about once a week,: when charged too highly, it is 

 not sufficiently retentive ; when not high enough, the instrument is 

 not sufficiently sensitive. With a charge of fifty units, which is my 

 average working charge, a difference of 1 in the angle of torsion is 

 equivalent to *1 of a unit of charge, so that if the difference between 

 air-reading and earth-reading is 1, the strength of atmospheric 

 electricity is 1 . The average strength of atmospheric electricity 

 observed has been from thirty to forty times greater than this, and 

 has in a few cases been 800 times greater. Hence it is obvious that 

 readings to whole degrees are abundantly sufficient ; and I have not 

 thought it necessary to aim at greater minuteness, more especially 



3c2 



