1306 



In the foregoing considerations we have assumed that a linear 

 relation holds between current and tension. From the fig. 2 and 3, 

 however, it appears that this it only true for a limited part of the 

 characteristics. The current-amplitude cannot rise above a definite 

 value. It follows that actually we have not to ask for the pure 

 imaginary roots of (27), but for the roots with a positive real part. 

 Hv an investigation as given above we only get to know the points, 

 where the real part of the roots p changes its sign, it appears to 

 nio that in most cases this will be sufïicient to discover tlic different 

 coupling-possibilities. 



in an American patent of Nov. 1917 (N". 102, 503) tiie use of 

 the andion as generatoi- is (lescril>ed, wiiere especially the quadratic 

 terms in rhe current-tension relation seem to assume the most 

 important part, it is rather difficult to draw conclusions from patent- 

 descriptions and as far as 1 know a discussion of such a coupling 

 has thus fai' not been pid)lished. 



