414 BELLINGER: HIGH FREQUENCY AMMETERS 



simultaneously. Thus the ratio of indicated current at high and 

 low frequency, for equal total current in the circuit, was obtained. 

 The high-frequency current was generated by the. oscillatory dis- 

 charge of a condenser across a spark gap. The instruments were 

 in a secondary circuit, loosely coupled to the primary, and con- 

 sisting of an inductance coil, the instruments, and Leyden jars, 

 in series. The chief source of accidental error in the observations 

 was the slight unsteadiness of the current together with the dif- 

 fering lag of the indicating devices of the instruments. 



From the dimensions and arrangement of the wires it was 

 found possible to predict quantitatively the changes of reading of 

 the wire instruments with frequency, while theoretical considera- 

 tions also made it possible to predict qualitatively the perform- 

 ance of the strip instruments. It is an interesting coincidence 

 that the changes of reading occur just in the range of radiotele- 

 graphic frequencies. In some of the instruments, the current 

 distribution was constant from low frequencies up to about 100,000 

 then underwent changes, and became constant for 1,500,000 

 and higher frequencies. Thus, in one sense, physically infinite 

 frequency was practically attained. The order of agreement of 

 calculation and experiment attained with these instruments is 

 in itself one of the most interesting results of the investigation. 

 It shows that calculations based on the formulas for self and 

 mutual inductance of finite portions of a circuit are valid, to the 

 observed degree of accuracy, for short lengths and for damped 

 high-frequency oscillations. 



In the case of the parallel wire ammeter, mutual inductances 

 between parts, which had hitherto been supposed to be negligible, 

 are the determining factor in the change of current distribution. 

 In the so-called unshunted ammeter, the errors are chiefly due 

 to the self-inductances of parts other than the hot wire. In this 

 case, they can be minimized by symmetrical location of the cur- 

 rent leads. In the strip ammeter, the terminal blocks have been 

 found to be the source of large errors. These are reduced by 

 proper shaping of the blocks. 



One effective means of reducing the errors of these ammeters 

 at high frequencies is by using working parts of high resistance. 



