106 



BELL SYSTEM TECHNICAL JOURNAL 



minima, where accordingly we may define Vsm and psm as before, 

 these do not always vary in the same sense with v. hs, the frequency 

 is raised from about 4- 10^ Vsm increases at first; then come the curves 

 with curious shapes; when again the fiattish minima return, at 4-108 

 cycles or thereabouts, Vsm is following the hitherto-familiar rule of 

 decreasing with increase of frequency; but further along, beyond 

 about 3 • 10^ the trend again reverses, and Vsm rises once more. There 

 is thus an "optimum frequency," at which (for a wide range of pres- 



1800 



5 1000 



uu OF Hg 



Fig. 19 — Onset-potential vs. pressure, in rarefied air, for self-sustaining glow 

 at the indicated wave-lengths, in tube described in context. (Gill & Donaldson, 

 Fhil. Mag.) 



sure) breakdown occurs at a lower voltage than when v is either lower 

 or higher. This comes at about 3-10^ cycles for tubes 10 to 20 cm. 

 long, low^er down for a 5-cm. tube. 



These complexities far surpass what other observers report. The 

 others, it is true, confined themselves to narrower ranges of frequency, 

 and yet their ranges were often so located on the frequency-scale that 

 they should have observed some of the striking reversals of trend and 

 distortions of curves, had the conditions been the same; to seemingly 



