296 Dr. Norman Campbell : Experiments on 



the slower is certainly not so much less than that of the more 

 rapid as the theory would indicate. Moreover, in all the 

 experiments on other machines which will be mentioned 

 later all the characteristic features of fig. 5 were reproduced. 

 It appears that whenever the damping is mainly due to losses 

 in the core, for a component of a given frequency the damping 

 is always much less if that component is the more rapid com- 

 ponent of an oscillation than if it is the slower component. 

 The result appears suggestive for any theory which would 

 take into account the effect of the losses due to hysteresis and 

 eddy-currents. 



Some experiments were made to discover how far the 

 damping of the oscillations was the same in the primary and 

 secondary circuits. More definite information on this point 

 will be given later, but it was clear that in general, and 

 especially for the longer periods, the damping in the primary 

 circuit was somewhat less than that in the secondary. It 

 was also suspected that the periods of the oscillations in the 

 two circuits were not quite equal, but the uncertainties in 

 the measurement of the period when the damping is so large 

 are too great (see p. 292) for any definite pronouncement on 

 this point to be made. 



Correction of the Peak Potential for Damping. 



8. It is clear from these results that the correction of the 

 calculated peak potentials for damping is not likely to improve 

 greatly their agreement with observation, if the calculated 

 values of the damping are used for the correction. In fact 

 these calculated values are so small that they would change 

 the calculated peak potentials by a few per cent, at most. 

 But it is worth while to inquire whether a better agreement 

 between the calculated and observed peak potentials can 

 be obtained by correcting the former for the observed 

 damping. 



In order to apply the correction strictly the values for both 

 the more rapid and the slower component ought to be taken 

 into account ; but the numerical work involved in applying 

 the correction in this manner would be extremely heavy and 

 certainly not justified in view of the uncertainties of the 

 value of the damping of the rapid component when its period 

 is of the same order as that of the slower. A simpler method 

 has therefore been adopted. According to the theory the 

 amplitude of the more rapid component should not be as 

 much as one-tenth of that of the slower for any value of Ci 

 shown in fig. 4 ; it has been simply assumed, therefore, that 

 the peak potential is determined wholly by the amplitude of 



