202 BELL SYSTEM TECHNICAL JOURNAL 



say, 500 ohms and the current through it adjusted to some convenient 

 value, such as 1 milliampere. This voltage of .5 volt will be sufficient 

 with most tubes to give a change in the plate current of 30 to 40 

 microamperes. 



The tul)e voltmeter is then reconnected to its normal place in the 

 circuit and the resistance R* is connected to the input of the amplifier. 

 Keeping the current constant at the value of 1 milliampere, the re- 

 sistance i?4 is adjusted until the change in the detector plate current 

 is the same as before. It is immediately apparent that the amplifica- 

 tion will be the ratio of the known voltage on the grid of the detector, 

 that is .5 volt, to the voltage on the input of the amplifier, as indi- 

 cated by the product of the resistance i?4 and the current through it. 

 The current having been kept constant, the amplification is the 

 quotient of the 500 ohms used when calibrating the detector ami the 

 resistance value obtained with the amplifier included. 



Considerable precaution must be observed to make sure that no 

 energy is getting into the amplifier circuit except that which ma\' be 

 measured by the voltage drop across the resistance Rt. This neces- 

 sitates the most careful shielding especialh' when the ani])hficati()n 

 is more than 50 times. 



With the measuring apparatus described a ilependable input \(>lt- 

 age as small as 1 millivolt can be obtained. The maximum amplifica- 

 tion which can be measured directly is, therefore, of the order of 500 

 limes when the output voltage to the detector is of tiie order of one 

 half of a volt. 



For the measurement of higher am|)litication liie following indirect 

 method may be used. 



The amplification is artificially decreased in some manner such as 

 reducing the number of stages in the circuit and this reduced amplifica- 

 tion is measured in the usual manner. The input current is then 

 reduced and the input resistance increased keeping the plate current 

 of the detector constant, the voltage impressed on its grid being 

 determined by the previous calibration. The amplification is now 

 increased to its normal value and the input resistance decreased until 

 the detector plate current has its original value. The ratio of decrease 

 in input resistance will thus give the increase in amiilification and the 

 total amplification will be the [jroduct of this and the smaller amplifica- 

 tion as first measured. 



The smaller current through the input resistance, which is oljtained 

 by this method and which will generally l)e less than can be determined 

 by the most sensitive thermocouple, will reduce the ])ick-up to a 

 sufficiently low value to give satisfactory results. In this connec- 



