MEASUREMENT OF INDUCTANCE 383 



used two steps for C?, and five steps for t]. It is possible by choosing 

 the correct values for ti to make the bridge direct reading for induc- 

 tance. The actual values used for the capacitance were .6 mf and 

 .06 mf. The values used for ri were 1,000/. 6 or 1,667 ohms and mul- 

 tiples or submultiples of this value. In this way the bridge was 

 made direct reading in millihenrys. 



The capacitance d has only one requirement to meet. It must be 

 small enough so that the ratio of resistance to reactance of arm AD 

 shall always be less than the ratio of reactance to resistance of the 

 coil. 



Taking 3,000 cycles as the maximum frequency, 10,000 ohms as 

 the maximum resistance in arm AD, and 200 as a maximum value 

 for the Q of the coil measured, then 



lirfC < 200, 

 and 



C < 1 mf. 



We have accordingly used a value of .6 mf in this arm to correspond 

 with the value of Cs- 



Resistances R and fo are dial type completely shielded resistance 

 boxes which can be varied from to 10,000 ohms in .01 ohm steps. 

 The resistances are all of the reversed layer type, wound on impreg- 

 nated wood spools and designed to give low phase angle and high 

 stability. 



The condensers are of the paraffine impregnated mica type, about 

 ten years old, thus ensuring high stability, and having temperature 

 coefficients less than .003 per cent per degree C, over the ordinary 

 range of working temperatures. 



The transformers are of a special type described by Shackelton.* 



Accuracy — Measurement of Inductance 



As previously stated the shielding, while increasing the stability of 

 the bridge, introduces capacitances across R and ri which increase 

 the corrections necessary in computing the effective resistance and 

 may also require corrections in the measurement of inductance if 

 sufficiently large. Accordingly, measurements were made on the 

 bridge to determine the magnitude of this error. By shunting R and 

 ri respectively, it was readily shown that capacitances as high as 

 200 mmf would not change the indicated inductance reading by as 

 much as .01 per cent for all settings of ri, for the whole range of R, 

 over the whole audio frequency range. This conclusion is in accord- 

 ance with equation 1. Since the shielding introduced capacitances 



