SF.I.F.CTIIF. CIRClirS .IXH SI.ITIC IXrr.RFF.RENCE 277 



// the sigmiling system requires the transmissions of a band of fre- 

 quencies corresponding to the interval uj — coi, and if the selective circuit 

 is efficiently designed to this end, then the mean square interference 



current is proportional to the frequency hand width 



This follows from the fad that, in tho c.isi- of ctlk-ifiitlj- designed 



liaiid-tiliors. (Ii>ii;iu<l to si-li'ct tlif frii|m'iicy ran^je "„ ~i- and 



2ir 



I'Mludf otliiT fri'(iui'niius, tlic inloi;ral / i-;^/-:— r,.; is proportional to 



Jo I Z (Joj) I' 



uij — uji to a high degree of approximation. 



The practical consequences of these propositions are imijortaiit and 



immediate. It follows that as the signalintj speed is increased, the 



amount of interference inevitably increases practically linearly and 



that this increase is inherent. Again it shows the advantage of 



single vs. double side-band transmission in carrier telephony, as 



fx)inted out by the writer in a recent paper.' It should be noted 



that the increased interference with increased signaling band width 



is not due to any failure of the selective circuit to exclude energy 



outside the signaling range, but to the inherent necessity of absorbing 



the interference energy lying inside this range. The only way in 



which the interference can be reduced, assuming an efficiently designed 



band lilter and a prescribed fre(iuenc\' range — =-r — -, is to select a 



2?r 



carrier frequency, at which the energy spectrum R (w) of the interfer- 

 ence is low. 



Formula (11) proi'ides the theoretical basis for an actual determination 

 of the static spectrum. Measurement of P over a sufficiently long 

 inter\al, together with the measured or calculated data for evaluating 



, _ . . , ,„ , determines R (ojm) and this determination 

 - i ^ ('") I" 



can be made as accurate as desired by employing a sufficiently sharply 



tuned circuit or a sulihciently narrow band filter. It is suggested 



that the experimental data could be gotten without great difficulty, 



and that the resulting information regarding the statistical frequenc\' 



distribution of static would be of large [practical value. 



The selective figure of merit S as defined by (14) is made up of two 



factors, which is inversely proportional to the re^iuired 



(0)2 — OJi) 



signaling frequency range; and the ratio of the integrals <r/p. This 



' Signal-to-Static-Interference Ratio in Radio Telephony, Proc. I. R. E. £., 

 June, 1923. 



