1076 THE BELL SYSTEM TECHNICAL JOURNAL, NOVEMBER 1952 



0.01 0.1 0.5 1 5 10 20 30 40 50 60 70 80 90 95 99 99.9 99.99 



PER CENT OF SAMPLES WHICH ARE SMALLER THAN THE ORDINATE 



Fig. 5 — Tj^pical distribution of test samples of r-f signal strength taken over a 

 small area. 



With the shape of the distribution known, only one other vakie need 

 be given in order to specify the propagation to such a small area. This 

 might be the median, the average, the rms, or any single point on the 

 curve. The one used most often here is the median, that is, the value 

 which is larger than 50 per cent of the samples and smaller than the other 

 50 per cent. Measurement of the median value by this statistical method 

 was found to be accurately reproducible, and therefore is presumed to be 

 reliable. Successive batches of 200 samples each, all covering the same 

 test area, yielded median values which differed not more than 0.5 db 

 when none of the conditions changed; i.e., transmitter power, antenna 

 gain, and receiver calibration remained the same. This accuracy may 

 seem surprising when it is realized that individual samples differ fre- 

 quently b}^ 10 db, and often as much as 30 to 40 db. 



It was presumed at the outset of the tests that the different frequencies 

 would exhibit different propagation trends with distance. For this reason 

 the samples have been grouped by distance. In presenting these results, 

 it was convenient to express the measurements of received RF signal in 

 terms of path losses. By this it is meant the loss between the input to a 

 dipole antenna at the transmitter and the output of a whip antenna on 

 the test car. These path losses will have, of course, the same distribution 

 as the received r-f signal. 



The results of the path loss measurements are given in Figs. 6, 7, and 

 8 for 150, 450, and 900 mc respectively. These values represent the loss 

 between the input to a half-wave dipole antenna at one end of the path 

 and the output of a quarter-wave whip at the other end. They are shown 

 here as a function of distance from the land station. For distances under 



