the Higli-Tension Magneto. 387 



interest, for, if it is successful, it will show a new method for 

 deriving the constants of the circuits and one that has some 

 advantages over methods which depend on the determination 

 of wave-forms. For the tracing of wave-forms of such high 

 frequency as is characteristic of magnetos and induction-coils 

 must always involve rather elaborate apparatus, whereas in 

 order to obtain (V2 m , Ci) curves only the simple apparatus 

 for measuring the peak potential is required (see page 301 

 of the March number) together with variable inductances 

 and capacities. 



In the deduction of the constants of the circuits from the 

 experimental curves, it is probably best to use only the 

 positions of the various maxima and minima, and to base no 

 conclusions on the relative height of these points ; for the 

 latter are likely to be more affected by slight imperfections 

 of the theory than the former. But even when the inquiry 

 is so limited, the best method of procedure requires a little 

 consideration. For it must be remembered that the experi- 

 mental curves give Y 2m as a function of Ci, while the theory 

 gives it as a function of u ; u is proportional to Ci, but the 

 factor of proportionality is not known at the outset ; to 

 discover this factor must be the first step in the process. 

 We may proceed thus : 



Each of the curves in fig. 13 applies to a different value 

 of L/ and therefore to a different value of the ratio w/Ci 

 and of c. The first step, then, will be to calculate from the 

 theory the values of w at which maxima and minima occur 

 for different values of c. According to Prof. Jones's theory 

 (see (40), (41) of his paper), maxima occur when u and c are 

 (very nearly) such that r = 4m — 1, and minima when they 

 are such that r = 4m + 1, m being any integer. In fig. 14 the 

 values of u at which minima occur for various values of m 

 are plotted against c ; fig. 15 gives the corresponding curves 

 for the maxima. In the latter figure the portions of the 

 curves which are drawn full instead of dotted are those for 

 which the correspondiug maximum is the highest maximum 

 of all ; the full lines, joined by vertical chain-dotted por- 

 tions, give the relation between c and that value of u which 

 corresponds to the optimum value of Cj (see § 10). 



From fig. 15 we see that the value of u which gives the 

 maximum m=l is very nearly independent of c within the 

 limits c = 0'2 to c = 0*4 and equal to 0*46. Accordingly, if we 

 can find the value of Ci, which gives the maximum ra = l, 

 we shall know that it corresponds to w = 0*46 and deduce 

 at once the ratio u/G 1 for the corresponding curve. The 

 maximum m = l will, if it occurs at all, be always that 



2E 2 



