the High- Tension Magneto. 299 



the damping which is due to hysteresis and eddy-currents, 

 and no theory which cannot predict this damping is of any 

 use in predicting the peak potentials. 



On the other hand, in some matters the theory gives 

 results which are in general agreement with observation. 

 Thus the relations between r and Ct or C 2 ' shown in fig. 2 

 are not very different from those predicted ; the values for c r 

 are approximately constant and not very different from the 

 value of c determined experimentally or the value to be 

 expected from the form and arrangement of the circuits. 

 Again, the changes in the inductances which result from 

 removing the armature from its housing are in the expected 

 direction, and the primary and secondary inductances are 

 reduced in approximately the same ratio (0*30 and 0*25 

 respectively); while removal from the housing is found to 

 reduce the coupling, as it should. Indeed there is only one 

 considerable discrepancy. When the armature is removed 

 from its housing, the secondary capacity C 2 should be 

 reduced ; but if we put L 2 = L 2 ' and calculate C 2 from T 2 , we 

 find C 2 = 30'8 mmf. in the housing and C 2 =51'5 mmf. out of 

 the housing, a difference which can hardly be accounted for by 

 an error introduced by identifying L 2 and L 2 ' '. It is probable 

 that the second value is the more nearly correct. 



It is not surprising that the success of the theory should 

 be so very different according as it is applied to the deduction 

 of the constants of the circuits, or to the prediction of the 

 peak potential. For in the former application we are only 

 concerned with the periods of the oscillations, and the general 

 theory of oscillations indicates that these periods will not be 

 very greatly affected by damping, so long as it not so great 

 as to reduce the oscillations to subsidences. On the other 

 hand, the prediction of the peak potentials involves the 

 amplitudes and the phase differences of the components of 

 the oscillations, and these will be greatly affected by the 

 presence of a source of damping which the theory does not 

 contemplate. 



Accordingly, in spite of the general failure of the theoiy 

 to predict absolute values for the peak potential, it is worth 

 while to inquire whether it can predict how those values vary 

 with alterations in the constants of the circuits ; for if it is 

 successful in this respect, its value in the design of machines 

 (the purpose to which the investigation was primarily directed) 

 will scarcely be diminished. 



The most remarkable and interesting feature of the theory 

 is the relation which it predicts between the peak potential, 

 the primary capacity C l3 and the coupling or the associated 



