22Q Prof. Perry on Long -Distance Telephony. 



resistance by the amount sl/k. But it is easy to *ee that if 

 we diminish self-induction or increase resistance we do harm 

 in telephony, and yet this kind of diminution through leakage 

 does good. On going into the matter carefully, it is seen 

 that it is the p being in the denominator of sr/kp 2 which 

 produces the good effect. In fact, if I and s are small, taking 

 p = 6000, p' = 600, we find 



Xal/^.{l-4246^-g^}. 



So that increasing s or I produces a good effect. Having 

 found the mathematical reason we have not far to go to find 

 the physical reason. 



It is evident from the tables that if we had no leakage we 

 could completely get rid of the evil effects of capacity by 

 introducing self-induction. It is also evident that if we 

 had no self-induction, we could completely get rid of the evil 

 effects of capacity by introducing leakage. But when there 

 is some leakage and some self-induction, we can in practice 

 only mitigate the evil effect of capacity ; for it is obvious that 

 although certain values of I and s give infinite distances, 

 doubling or halving these values produces enormous diminu- 

 tion in distance, and such a constant of a cable as s may alter 

 very greatly. 



About fifteen years ago, with Prof. Ayrton I made many 

 experiments on signalling through bare copper wires lying at 

 the bottom of the water in the moat of Yedo in Japan. Here 

 k and s were both very great. We had much less success 

 than we expected, and we abandoned, perhaps too readily, 

 our idea of a very cheap submarine cable. 



The following tables are of general application. Let the 

 numbers given in Table III. be divided by the value of Vkr 

 for any cable or conductor of a telephonic line, and let them 

 also be divided by the value of m which is considered suitable*, 

 and they will become the limiting distances X in centimetres 



I s 



for that conductor, for the various values of - and - given. 



Let the numbers in Table IV. be divided by the value of 



sjkr for any conductor of a telephonic line, and let them also 



be divided by the value of n which is considered suitable, and 



they will become the limiting distances Y in centimetres for 



Is 

 that conductor, for the various values of - and T given. 



r k & 



* It is more correct to say that the numbers are to be multiplied by 

 log e (l+- 



