CONTEMPORARY ADVANCES IN PHYSICS 107 



minor dlfiferences in the conditif)ns, then, the discrepancies must be 

 ascribed. The complexities are not peculiar to hydrogen, for Gutton 

 obtained a very similar set of curves with oxygen, and in much earlier 

 work (1923) on rarefied air he found F.sm increasing with frequency 

 up to about 7.5- 10'^ cycles, and thenceforward diminishing all the way 

 to the uppermost limit of his frequency-range, 2.14-10^. This last- 

 mentioned result was obtained with an external-electrode tube, the 

 exterior tinfoil belts 24 mm. apart; on substituting an internal- 

 electrode tube, he found W,,, increasing with v over the entire fre- 

 quency-range. But I must leave the reader to explore and collate 

 Gutton's numerous curves for himself, and mention only in closing 

 that in a tube of rarefied hydrogen with external electrodes 53 mm. 

 apart, he got at frequency 2.5-10^ a breakdown-potential of only 57 

 volts — an amazingly small value, far lower than anything ever obtained 

 with direct current. 



Gill and Donaldson produced Vg-vs-p curves with two minima 

 apiece instead of one, by placing the long slender discharge-tube 

 (20 cm. long, 2>.2> cm. diameter) between two metal plates serving as 

 the electrodes, with its axis parallel to their planes. These curves 

 were obtained in rarefied air, with various frequencies between 3.5- 10^ 

 and 2.^ • 10^ corresponding to wave-lengths between 86 and 125 meters; 

 one sees them in Fig. 19. (Below and to the left are curves for four 

 other and higher frequencies, ranging from 5 • 10^ to 1.5 • 10'^; these have 

 the familiar single-minimum controur, and both Vsm and psm decrease 

 as V increases.) Thereupon, Gill and Donaldson re-oriented the tube 

 so that its axis was perpendicular to the electrode-plates — owing to 

 its length, it had to be passed through a pair of holes made specially 

 in the plates — and repeated the observations. Now, of the two 

 minima, the one to the right disappeared ; for each of the several wave- 

 lengths, the curve continued straight on past the point marked D in 

 Fig. 8, to a single minimum lying far to the left. 



Maintaining-Potentials of High-Frequency Glows in 

 Rarefied Gases 

 When the high-frequency glow in a rarefied gas is established, the 

 voltage between the electrodes — that is to say, the amplitude V of 

 the oscillating voltage — is as a rule much smaller than the breakdown- 

 potential. It would seem natural to begin the study of the glow by 

 determining the curves of current versus voltage and current versus 

 length {i.e. anode-to-cathode distance) for many values of pressure, 

 as the custom is in dealing with direct-current discharges; but data of 

 this sort are few. Further along I will speak of work of Townsend's 



