77. A . Rowland — Electrical Measurement- 433 



Mr. Penniman will continue the investigation with other 

 condensers, liquid and solid, as well as plates in electrolytic 

 liquids. 



The results in the other measurements have been fairly satis- 

 factory, but many of the better methods have only been 

 recently discovered and are thus untried. But we must ac- 

 knowledge at once that work of the nature here described is 

 most liable to error. Every alternating current has, not only 

 its fundamental period, but also its harmonics, so that very 

 accurate absolute values are almost impossible to be obtained 

 without great care. To eliminate them, I propose to use an 

 arrangement of two parallel circuits, one containing a conden- 

 ser and the other a self-inductance, each with very little resist- 

 ance. The long period waves will pass through the second 

 side and the short ones through the condenser side. By shunt- 

 ing off some of the current from the second side, it will be 

 more free from harmonics than the first one. 



However, in a multipolar dynamo, especially one containing 

 iron, there is danger of long period waves also, which this 

 method might intensify. A second arrangement, using the 

 condenser side, might eliminate them. However, many dyna- 

 mos without iron and without too many poles and properly 

 wound produce a very good curve without harmonics, especially 

 if the resistance in the circuit is replaced by a self inductance 

 having no iron. These remarks apply only to absolute deter- 

 minations. Ratios of inductance, self and mutual, and capacity 

 are independent of the period, and thus it can always be elimi- 

 nated. Measurements of resistances also are independent. 



But there are other errors which one who has worked with 

 continuous currents may fall into. Nearly all alternating cur- 

 rents generate electromagnetic waves which are so strong that 

 currents exist in every closed circuit with any opening between 

 conductors in the vicinity. 



We eliminate this source of error by twisting wires together 

 and other expedients. But in avoiding one error, we plunge 

 into another. For, by twisting wires we introduce electro- 

 static capacity between them, which may vitiate our results. 

 Thus, in methods 23 or 24 for comparing mutual inductances, 

 if there is electrostatic capacity between the wires, a current 

 will flow through the electrodynamometer in the testing cir- 

 cuit and destroy the balance. 



Various expedients suggest themselves to eliminate this 

 trouble, as, for instance, the variation of the resistance A in 

 the above, but I shall reserve them for a future paper. I may 

 say, however, that it is sometimes possible, as in method 12 for 

 instance, to choose a method in which the error does not exist. 



