382 Dr. Norman Campbell : Experiments on 



values of c. In the Appendix tables are given showing the 

 numerical values of Y 2 m and also of r from which the curves 

 are plotted ; these tables may possibly be of use to others 

 pursuing similar investigations. When all constants but Cx 

 are unchanged, u is proportional to Cj ; and accordingly 

 the experimental curves of fig. 9 should each resemble 

 one of the theoretical curves of fig. 10, or a curve inter- 

 mediate between those drawn for some intermediate value 

 of c*. 



It is evident at once that the theory is not applicable. 

 In the experimental curves there is no trace of the sharply- 

 defined minima which are so characteristic a feature of the 

 theoretical curves. Minima can be traced in the experi- 

 mental curves, and they appeared always to be accompanied 

 by a variability in their neighbourhood greater than the 

 average; but the minima are always shallow depressions 

 and not sharp cusps. Further, if we neglect the subsidiary 

 minima and consider only the general trend of the curves 

 no better agreement is found. All the theoretical curves 

 have main maxima at values of u which are markedly 

 different from 0, and the value of u at which the maximum 

 occurs varies notably with c. But all the experimental 

 curves have maxima at or very near 0, and there is no sign 

 of a variation of the position of the maximum with c. 



In particular, the expectation in which the experiments 

 were undertaken is completely falsified. There is no sign 

 that by increasing c and decreasing the coupling the 

 optimum value of Cx can be increased, or even that the 

 decrease of the peak potential with increase of Oj can 

 be diminished. Indeed, if the curves are compared for 

 two combinations which have the same number of coupled 

 turns but different numbers of uncoupled turns {e. g. b and c 

 or c and A), it will be seen that the addition of uncoupled 

 turns, which decreases the coupling, always makes the curve 

 fall off more steeply as Ci is increased. 



And when we turn our attention from the general shape 

 of the curves, and compare merely the maximum peak 

 potentials given by different combinations with the optimum 

 value of Cx in each case, no better agreement is found. 

 If hi and G 2 are constant (as they are very approximately 

 in these experiments) then the theory predicts that the 

 relation between the peak potential with optimum Cj 



* The abscissae are so chosen that the scale of u in fig. 10 is nearly the 

 same as the scale of C-l in fig. 9. Differences between the two figures are 

 not simply due to a wide difference in the scale of the abscissae. 



