October 20, 1916] 



SCIENCE 



577 



oscillator sold by the Western Electric Com- 

 pany, as it gives a pure sine wave form and the 

 frequency of the current can be varied at will 

 over a range of 160 to 4,200 cycles. We have 

 also arranged with Vreeland for Leeds and 

 Northrup to sell a smaller instrument giving 

 500, 750, 1,000 and 1,500 cycles per second. 



Curtis coils were used in our bridge, as by 

 this means we could practically eliminate in- 

 ductance and capacity effects which are a 

 source of error when inductive resistances are 

 used. Tuned telephones sold by the Western 

 Electric Company were used so that the mini- 

 mum could be detected as accurately as pos- 

 sible. 



Several different bridge arrangements have 

 been used for measuring the capacities, induct- 

 ances and resistances. In order to measure a 

 " cell constant " it is necessary to determine 

 both the resistance and the capacity given by 

 the standard solution of (N/10, N/50, 

 iV/1,000) potassium chloride at different fre- 

 quencies. A substitution method suggested 

 by Curtis for measuring resistances was used, 

 as it prevents errors due to any changes in the 

 resistances of the bridge coils, or of the in- 

 ductance, caused by variations in temperature, 

 and allows the experimenter to read 'the re- 

 sistances directly instead of having to make 

 involved calculations. 



Resistance measurements made on a given 

 solution in a given cell by using (1) an in- 

 ductance in series with the cell and (2) a con- 

 denser in parallel with the resistance arm 

 checked to within 0.001 per cent, when all nec- 

 essary corrections are made. 



We have used a new type of cell in which 

 each electrode is supported in four places, so 

 that it can not move and thus change the 

 cell constant. We have also made all joints 

 to come below the surface of the water so that 

 there can be no error from evaporation of the 

 solvent. One criterion of good cells is that 

 whatever the solutions used the ratios of the 

 resistance of any solution in two such cells, or 

 of two solutions in any cell, must be constant 

 to within 0.01 per cent, at infinite frequency. 



It was found that resistance readings on a 

 given cell with a given solution can be re- 



peated easily with an average deviation within 

 ± 0.001 per cent. 



Even in the old type cells, and when no 

 special precautions were taken in transferring 

 the solutions, resistance measurements on dif- 

 ferent parts of the same solution in the same 

 cell checked to within 0.01 per cent. 



When solutions of electrolytes were allowed 

 to stand in cells for 24 hours, the resistance of 

 the solution in the cells with platinized elec- 

 trodes did not change, while in those with 

 bright electrodes the resistance increased at 

 the rate of about 0.003 ohm per hour. 



There is no measurable change in the resist- 

 ance of a solution, or the capacity of the cell, 

 with change in voltage from 0.25 to 8 volts, 

 provided the cells, solutions and containers are 

 kept scrupulously clean, and the polarization 

 voltage is kept below 1.23 volts. Dr. G. H. 

 Gray is studying this problem by the use of 

 the oscillograph. 



The capacities in cells with bright electrodes 

 vary from 10 to 1,000 microfarads, while in 

 cells with platinized electrodes they range 

 from 500 to 5,000 microfarads. 



There is no measurable change in the resist- 

 ance of a solution with change in frequency 

 from 600 to 2,000 in our cells with platinized 

 electrodes 1 inch in diameter, and such cells 

 therefore give approximately the true elec- 

 trical resistance of solutions at any such fre- 

 quency. The ratios of the resistances of N/5, 

 N/10 and i^/20 Nad measured in two differ- 

 ent cells with platinized electrodes show a 

 deviation of not over 0.01 per cent, at these 

 frequencies, which is hardly more than the 

 experimental error. 



In cells with bright platinum electrodes 

 there is a change in resistance with change in 

 frequency from 250 to 3,000 cycles and higher, 

 and this change depends upon several factors: 

 (1) As the concentration of any given solu- 

 tion is decreased, and therefore the resistance 

 increased, the change in resistance with change 

 in frequency is decreased; (2) as the area of 

 the electrode surface is increased the change 

 in resistance with change in frequency is de- 

 creased; (3) as the area of the electrode sur- 

 face is increased the inductance necessary to 



