Potts — Rowland's New Method. 



107 



compared to the first for any values of h used in this work. 

 The second term will in the most unfavorable circumstances 

 amount to only 1 part in 10,000, so that the corrections due to 

 the electrostatic action of the coil may be entirely neglected. 



The next point investigated was the correction due to the 

 electric absorption. In these observations the absorption was 

 determined, the capacity then measured, and the electric 

 absorption again determined. In Table IX are given the 











Table IX. 













r- 



=4811 



R2 = 



= 707-8 



1X4- 



L 

 = 202-G 



= 5-302 in 2 and 

 5-318 in rest. 



Date, etc. 





R3 



Bi 



Calculated 

 R'i 



A 



(R'i— R 



) T 



A 

 T 



Mean. 

 A 

 T 





1 



2195- 



34-90 



65-35 



30-45 



•117 



261- 





N-16 

















253- 





7 



2614- 



34-97 



54-88 



19-91 



•0816 



244- 









R« 



R s 



T 



A 



Corrected for 

 slectric absorption. 



Ra Rft 



106x ^k =c 





2 



238-2 



2604 



•121 



30-6 



268-8 



2604 



7574 M.F. 





3 



228-8 



3053 



•103 



26-1 





3079- 



7-549 









4 

 5 



328-2 

 431-5 



2119 

 1602 



•103 



•102 



26-1 



25-8 





2145- 

 1628' 



7-554 

 7-569 













6 



531-5 



1302 



•0908 



23-0 





1325- 



7-552 



results with the condenser in one arm and then changing to the 

 other, and also changing the resistance in series with the con- 

 denser. The results are corrected for electric absorption and 

 the change in L due to the change of c and the coil from one 

 arm to the other, caused by the coils of the electro-dynamome- 

 ter having different coefficients of induction. The greatest 

 difference between two determinations under these different 

 conditions is 3 parts in 1,000. 



The change of capacity with the period of the current was 

 now tried. Table X shows the results of the investigation. 

 The error of each observation has a limit of about 1 part in 

 1,000, if the observations are compared among themselves, 



while the actual error, as compared with the true ratio -^ may 



be in error two or three times this ; but we are not particularly 

 concerned here with the actual value, but merely the change 

 with change of period. In fig. 7 the results are plotted on two 

 scales. The results for the capacity seem to agree very well 



Am. Jour. Sci.— Fourth Series, Vol X, No. 56. — August, 1900. 



