MEASUREMENTS AT TELEPHONIC FREQUENCIES. 117 



very distinctly greater. Consequently, when using this subdivided 

 resistance as a volt box, the B coils give too small, and the A coils 

 too large a result. 



It may be observed that it is impossible completely to avoid the 

 introduction of these virtual shunting and leak condensers, in the 

 construction of hi8:h-resistance boxes for a.-c. tests. If these vir- 



FiG. II. Exploration, by means of the potentiometer P, of the voltage drop 

 along a high-resistance box AB, of 100,000 ohms. 



tual condensers are dissymmetrical, and vary from coil to coil, it 

 becomes very difficult to make a proper correction for the error due 

 to charging currents. If, however, care is taken to preserve sym- 

 metry of structure throughout the resistance box, so that the distri- 

 butions of virtual capacitance remain uniform from coil to coil, the 

 correction may be readily computed, at any impressed frequency, by 

 hyperbolic functions, according to the well-known theory of arti- 

 ficial lines, from a single set of observations for determining the line 

 constants. A knowledge of the error will thus virtually eliminate 

 the error. Suitably supported and connected metallic ground shields^ 

 surrounding each resistance coil, would serve to distribute the leak 

 capacitance uniformly. High-resistance section boxes should, there- 

 fore, be designed so as to subtend small hyperbolic angles per sec- 

 tion ; but whatever the section angle is, it should be regular 

 throughout. 



