1200 THE BELL SYSTEM TECHNICAL JOURNAL, SEPTEMBER, 1953 



0.075 per cent of the time, and (2) that portion of Curve H (on Fig. 8) 

 which corresponds to fades of less than 30 db for the remaining 99.925 per 

 cent of the time. Curve / on Fig. 9 is such a curve. 



If frequencies separated from the assumed operating frequency by 80 

 mc were used instead of those assumed above. Curve C instead of 

 Curve B on Fig. 7 would have been used. This would have shown a 

 somewhat smaller improvement because of some correlation between 

 the fading on the assumed operating frequency and an alternate fre- 

 quency separated from it by 80 mc. However, the improvement gained 

 from the use of this system would be ample to prove-in its use. 



The question is sometimes raised concerning the reason that the 

 curves on Fig. 9 show an apparent improvement for fading of less than 

 30 db, since the circuit is switched only when fading deeper than 30 db 

 occurs. This is because the curves are cumulative distribution curves; 



-* 4 8 12 16 20 24 28 32 36 40 44 48 



PATH LOSS IN DECIBELS (RELATIVE TO MIDDAY NORMAL) 



Fig. 8 - Statistical (nsfribution of fading loss Princeton-Lowden, Iowa, path 

 July and August, 1950. 



