68 Mr. W. Gannon. 



the framework with the three plates. In addition to the voltameter, 

 the jar contained a small mercury gauge. The upper rim of the jar 

 was ground, and previous to exhaustion was coated with grease, 

 upon which pressed a ground-glass plate. The current was derived 

 from storage cells, and was measured by a tangent galvanometer or a 

 Thomson's magnetostatic centiamperemeter ; it was adjusted to the 

 proper value, and kept constant throughout an individual experiment 

 by having in circuit a carbon rheostat and a set of specially prepared 

 resistances of different values. The copper plates were made of 

 commercial sheet copper (24 B.W.G.) : they were cleaned with 

 emery paper, nitric acid, caustic potash, tap water, and, finally, dis- 

 tilled water, and a deposit was laid on them before being weighed 

 for an actual experiment. The deposit was treated in the way 

 recommended by Gray ; I found it advantageous to have the distilled 

 water hot, as on removing the plate the deposit dries more quickly, 

 and consequently the possibility of oxidation occurring is diminished. 

 The plates, after being in the balance case for a variable period 

 never less than three hours, generally over night were weighed on a 

 16-in. Oertling balance, which weighed accurately to 1 tenth-milli- 

 gram. The solutions were made by dissolving ordinary commercial 

 sulphate in distilled water and filtering ; the solutions with no fret- 

 acid added were used immediately, so as to conduct the electrolysis 

 before the formation of basic salts ; the acid solutions contained 1 per 

 cent, free sulphuric acid. When all the joints were air-tight, by 

 being closed with cement, no difficulty was experienced in retaining 

 the partial vacuum constant throughout each individual experiment. 

 Leakage of the current at either voltameter was tested for throughout 

 the experiments, and this point was considered satisfactory when, 

 both voltameters being under atmospheric pressure, the deposits did 

 not differ by more than 1 or 2 tenth-milligrams. Particular atten- 

 tion was paid to the handling of the plates before and after each 

 experiment, especially in the replacing in, or removal from, their 

 clips. By means of a small rod, acting as a lever, friction was pre- 

 vented between the clips and plates, and hence the slight danger of 

 loss of copper prevented. Immediately after the current was 

 stopped in each experiment, the framework holding the plates was 

 quickly removed from the solution and plunged into distilled water 

 so as to remove all solution from the plates oxidation occurs very 

 rapidly if any solution remains on the plates, even for a very short 

 time. 



My first object was to ascertain if any difference in the weight of 

 the deposit was caused by placing one voltameter in vacua, without 

 any special regard to the current density (which throughout all the 

 experiments was the same at both cathodes). The experiments 

 arranged in Table I show an appreciable, though inconstant, differ- 



