the High- Tension Magneto. 391 



L/ = '00200 lies far from any straight line that can be drawn 

 through the remainder ; but if we neglect this point, we 

 get the line B, giving L 1 = -00283 and L 12 L 21 /L/ = '00270. 

 Ijj is again smaller than the value determined by the oscil- 

 lation measurements, namely '00578 — 00100 = '00478. 

 Using the line to recalculate c we obtain the figures shown 

 under c (calc.) ; it should be noted that for IV = 0, c (calc.) 

 is 0*046, agreeing well with the observed value by the oscil- 

 lation method 0'040. 



We may now return to Table III., and, using this value of 

 L 1? calculate T 2 from u/C } . We obtain for the various values 

 of IV, T 2 = -000192, -000182, '000194, -000188, '000182. 

 T 2 determined by the oscillation method is '000258 sec. 



It is clear, then, that the values obtained from the (V 2m , Cj) 

 curves are not perfectly consistent amongst themselves, and 

 that different methods of calculation give somewhat different 

 results ; further, that these values do not agree with those 

 found by the oscillation measurements. On the other hand, 

 we obtain values of the right order of magnitude, and it is 

 possible that the method, if applied to an induction-coil, for 

 which the theory appears much more nearly correct, might 

 give the constants with as great accuracy and much less 

 trouble than by measurement of the periods. To complete 

 the determination L 2 ' would be required. If the variation 

 of the peak potential with C 2 , instead of Ci, were studied, a 

 simple extension of the method would give the required 

 information ; but it has not been thought worth while to 

 pursue the matter further. 



One final point may be investigated. Owing to the 

 uncertainty in T 2 it is not worth while to attempt to com- 

 pare measured and calculated values for the peak potentials, 

 but we may investigate how the measured values vary 

 with L/. According to the theory, the peak potential ought 

 to be nearly proportional to v/Lx + L/; applying this 

 criterion we obtain the following table according as we 

 take for Lj the value found from the determination of the 

 periods or the method just given : — 





Table V. 





L/. 



L,= '00478. 



L T =-00283. 







3-7 X10* 



4-82 xlO 4 



■00025 



3-54 



4-53 



•00050 



3-83 



4-57 



•00075 



3-70 



4-61 



•00100 



369 



4-54 



►00200 



3-76 



4-45 



•00400 



3-78 



4-27 



