A NEW CURRENT WKKUIKK, ETC. 527 



weights placed in position, and, at a given signal, the balance coils were included in 

 the circuit. The resistance in N, fig. 23, was rapidly adjusted until (1) the ammeter 

 reading appeared to he the same as hefore, (2) balance waa obtained when S' was in 

 the potentiometer circuit, and (3) the fulfilment of the latter condition when 8 wan 

 substituted for S'. In general, these adjustments occupied about 10 seconds. When 

 '"in li tii u i (2) held, a signal was made to the balance operator, and the beam of the 

 balance was freed. The average duration of a complete set of observations was 

 -iJ minutes, and during this time the balance coils were included in the circuit for 

 about 12 minutes. 



General Behaviour of the Brilance. After eliminating the difficulties mentioned 

 on pp. 523-520, the working of the balance, when cold, was most satisfactory. Under 

 normal conditions the constancy of the rest-point of the balance is well within O'l 

 scale division when no current passes through the coils, and the sensitiveness is about 

 8 divisions for 10 milligrammes. When a current, passes through the coils for not 

 more than 20 minutes the same constancy is in general maintained, and if the 

 balance circuit is occasionally broken as it is in experiments for the determination 

 of current this interval of constancy is prolonged to 30 minutes or more. If the 

 current through the balance coils is maintained after this interval, approximating to 

 30 minutes, the balance Incomes unsteady, and no very accurate observations can be 

 made ; if, however, the circuit is broken after the interval, the balance reading remains 

 approximately constant, variations of the order of 0'2 scale division only being 

 observed. At the end of three or four hours another determination of current is 

 possible, with practically the same degree of accuracy as before, but soon after these 

 observations the balance becomes unsteady, and shows variations in the rest-point, 

 gradually increasing from O'l to I'D scale division. If the second set of observations 

 are made within one or two hours of the first set, the balance reading is not constant, 

 and the results obtained are not of a high order of accuracy. In general, therefore, 

 only two determinations of current are possible within six hours, but these are 

 associated with a very small observational error. One determination normally 

 occupies from 16 to 25 minutes. 



Our usual procedure was to make one complete set of observations in the morning 

 and another in the afternoon, after the balance had been cooling for several hours. 

 Attempts made on several days to make a third set were never successful. 



The time which elapsed tat ween morning and afternoon observations of E.M.F. 

 was usually devoted to silver-deposit determinations, the standard cell S and 

 resistance R, fig. 23, l>eing used for keeping the current steady at a calculable value 

 during the deposition. In effect, therefore, the combination of cell and coil, forming 

 a secondary standard of current, was standardised morning and afternoon by the 

 iNilance, and used in the interval for measuring the current through the voltameters. 

 As, however, the determination of the electro-chemical equivalent of silver forms the 

 subject of another paper, it need not be discussed here. 



