290 BRJ.L SYSTEM TECHNICAL JOURNAL 



signal of strength 380 iiv/m in the loop.^ The point C on the curve 

 shows that at 8:25 A.M. the static intensity received on a 2000 cycle 

 wide frequency band corresponded to the energy received from a c.w. 

 signal of strength 3.8 ixv/m. It would be practical always to relate 

 static to such a c.w. signal. Experiments arc now being conducted 

 to determine whether the energy received from static is proportional 

 to the width of the frequency band of the receiving set and if such is 

 found to be the case then it is proposed to have the data relate to a 

 1000 cycle wide band. That static is, say, 7 microvolts per meter per 

 kilocycle (7 ixv/mkc) would then mean that the energy of the static 

 received on a 1000 cycle wide frequency band is the same as the energy 

 received from a c.w. signal of strength 7 nv/m. 



Attempts have been made to calibrate the set by inducing in the 

 loop, voltages of the shape shown in Fig. 5. Relating static to such 

 signals would have the advantage of being independent of the band 



to 



o 



> 



TIME 

 Fig. 5 — Shape of impulse voltage 



width of the set. Such signals were obtained by closing and opening 

 a mercury switch, but one signal per second, or 10 impulses per period, 

 would overload the set (the tubes) very much. At least 10 impulses 

 per second would be required if the set should not be overloaded by 

 each individual impulse, but this would be a difficult task to accom- 

 plish and it is therefore recommended that static be measured as 

 explained above, by inducing a local c.w. singal into the loop. The 

 fact that five static crashes in the course of 10 seconds — one period — 

 does not overload the set while 100 impulses of the shape shown in 

 Fig. 5 are required to prevent overloading gives us some interesting 

 information on static. It shows that a single static crash is not a 

 single sudden change of the field in the ether and that it cannot be 

 represented by less than 20 consecutive impulses. 



The record of Fig. 4 shows that each step on the gain control poten- 

 tiometer is 4 TU and the selection of such steps and of 15 seconds will 

 now be discussed. To decrease the 4: TU step to a 1 TU step would 



* It may be worth while to have such a calibration signal introduced automatically 

 for instance once every two hours. 



