612 BELL SYSTEM TECHNICAL JOURNAL 



and setting 2\ = 2900 degrees Kelvin 



F = 1.9 for F = 10 

 F = 1.09 for F = 100 



Assuming that F can be read to within ±1% our accuracy in determining 

 F would be within about ±1% for F = 10 but only within about ±10% 

 for F = 100. If the noise source had a temperature of 40 X 290 degrees, our 

 experimental errors would be reduced accordingly to about ±1/4% for 

 F = 10 and ±2.5% for F = 100. Since metal wires will not stand such tem- 

 peratures, we must look to something different for the noise source if these 

 accuracies are to be achieved. 



In view of the foregoing considerations, the nonoscillating reflex klystron 

 presented one possibility of a suitable microwave noise source. This, how- 

 ever, was not exploited because the bandwidth was not wide enough. 



Another possibility was found to be an electrical gas discharge. This type 

 of source was determined to generate noise at microwave lengths when the 

 open end of the input-waveguide of a sensitive microwave receiver was 

 directed toward various gaseous discharge tubes, including a 721 A TR 

 tube containing water vapor and hydrogen, a neon light m a stroboscope, 

 a mercury vapor rectifier and an ordinary fluorescent desk lamp. Of .these, 

 the commercial fluorescent lamp appeared to lend itself most readily to 

 mounting in a waveguide without the complication of the effects of the 

 internal metal electrodes, so further tests were performed on it. 



Microwave Measurements 



A T-5, 6-watt, daylight fluorescent lamp,^ lighted from a d-c. source, 

 was mounted with its axis parallel to the magnetic vector in a waveguide 

 as illustrated in Fig. 2. The lamp itself was 9" long, with cathodes at each 

 end. These could be isolated from the field in the 1" x 2" waveguide by 

 enclosing the portion of the lamp which extended beyond the walls of the 

 waveguide in cylindrical metal shields which formed waveguides beyond 

 cutoff. Tlius, energy was kept from reaching the cathodes, and the noise 

 source was effectively confined to that part of the discharge which appeared 

 inside the main waveguide. A piston in back of the gaseous discharge tube 

 served to tune out the susceptance and a trimming screw provided an 

 additional adjustment. The conductance could be adjusted by varying 

 the direct current. 



The admittance of the combination could be adjusted for an impedance 



^ A commercial tluorescciU lam]) contains about two mm. of argon and si.x to ten microns 

 of mercury gas. The argon merely facilitates the initiation of the discharge; the mercury 

 furnishes the radiation which excites the fluorescent material. 



