PART I — THE GROWING NOISE PHENOMENON 839 



(2) The lower the frequency, the lower the field strengths at which 

 noise amplitudes change most violently with field. 



(3) The three noise curves resemble each other in small details. 

 The results of a great many records of the kind illustrated by Figs. 



2 and 3 can be summarized as follows: 



(1) There is always a decrease in beam-ripple amplitude associated 

 with noise growth at any frequency. (Sometimes the ripple amplitude 

 increases afterwards, as in Fig. 3.) 



(2) The higher the frequency, for a given field, the more articulated 

 or scalloped the noise pattern. 



(3) No correlation can be found between rate of noise growth and 

 either (a) distance from gun to take-off plane, or (b) net gain at the end 

 of the drift region. The trends, as a function of magnetic field, are differ- 

 ent at different frequencies. 



(4) Greatest noise growth does not, as a rule, occur with zero flux 

 threading the cathode. Sometimes two nearly equal peaks occur for two 

 values of Be , each of opposite polarity, referred to the sense of the main 

 field. 



(5) The noise-distance patterns change very slowly with frequency. 



(6) No beam entirely ripple-free throughout its length has ever been 

 observed by the writer. 



IV ORIGIN OF GROWING NOISE 



If noise growth is due to some amplification process, it should be 

 possible to adjust the beam-focusing conditions so that the noise currents 

 start increasing at the anode, and attain the greatest possible over-all 

 gain at the end of the drift space. The enhanced activity of the unknown 

 gain mechanism should presumably help identify it. The curves of Fig. 

 4 show that maximum noise occurs at different values of the focusing 

 field, for different values of field at the cathode, and different probe posi- 

 tions. With the anode voltage and receiver freciuency fixed, therefore, 

 the conditions for greatest net noise growth can only be found by a series 

 of trial settings of both magnetic fields, each followed by a recording of 

 the noise-distance pattern. Eventually, a set of fields can be found for 

 which the greatest total gain occurs; and such patterns are usually found 

 to show fairly steady noise-amplitude increase, on the average, over the 

 entire length of probe travel. 



The results of this procedure for noise power near 4 kmc, as well as 

 the patterns of collector-current versus distance with the same fields, 

 using the 0.1 00-, 0.050-, and 0.020-inch apertures fixed at the probe 

 centerline, are shown in Fig. 5. A similar set of records, for noise power 



