H. A. jRowland — Electrical Measurement. 447 



12, results to 1 in 1000 can be obtained. Replacing 1/ by an 

 equal coil, the ratio of the two, all other errors being eliminated, 

 can be obtained to 1 in 10,000, or even more accurately. 



The main error to be guarded against in method 12, or any 

 other where large inductances or resistances are included, 

 arises from twisting the wires leading to these. The electro- 

 static action of the leads, or the twisted wire coils of an 

 ordinary resistance box, may cause errors of several per cent. 

 Using short small wire leads far apart, the error becomes very 

 small. 



Method 6 is also very accurate, but the electric absorption 

 of the condensers makes much accuracy impossible unless a 

 series of experiments is made to determine the apparent resist- 

 ance due to this cause. 



In method 12 I have not yet detected any error due to 

 twisting the wires of coils I. However, the electrostatic action 

 of twisted wire coils is immense and the warning against their 

 use which I have given above has been well substantiated by 

 experiment. Only in case of low resistances and low induct- 

 ances or in cases like that just mentioned is it to be tolerated 

 for a moment. Connecting two twisted wires in a coil in 

 series with a resistance between them, I have almost neutralized 

 the self inductance, which was one henry for each coil or four 

 henrys for them in series ! 



Altogether the results of experiment justify me in claiming 

 that these methods will take a prominent place in electrical 

 measurement especially where fluid resistances, inductances 

 and capacities are to be measured. They also seem to me to 

 settle the question as to standard inductances or capacities, as 

 inductances have a real constant which can now be compared 

 to 1 in 10,000, at least. 



The new method of measuring liquid resistances with alter- 

 nating currents allows a tube of quite pure water a meter long 

 and 6 m,n diameter having a resistance of 10,000,000 ohms to be 

 determined to 1 in 1,000 or even 1 in 10,000. The current 

 passing through the w^ater is very small, being at least 500 

 times less than that required when the bridge is used in the 

 ordinary way. Hence polarization scarcely enters at all. 



It is to be noted that all the methods 15 to 24 can be modi- 

 fied by passing the main current through one coil of the 

 electrodynamometer and the branch current through the other. 

 The deflection will then be zero for a more complicated 

 relation than the ones given. If, however, one adjustment is 

 known and made, the method gives the other equation. 



Thus method 18 requires H / R" — R'R^O. Hence, when 

 this is satisfied we must have the other condition alone to be 



Am. Jour. Sci.— Fourth Series, Vol. IV, No. 24.— Dec, 1897. 

 31 



