SOME CONTEMPORARY ADVANCES IN I'llYSICS—XI 471 



length Xo lying between 504A (helium) and 3184A (caesium); and this 

 is the most troublesome region of the spectrum to deal with, partly be- 

 cause light of wavelengths King within it is tremendously absorbed by 

 nearly all solids and even gases, and partly because good sources for 

 such light are difficult or impossible to procure. Even in the compara- 

 tively accessible zone between 2000A and 3500A it is customary to use 

 the light of the mercury arc, which provides a few widely-spaced bright 

 spectrum-lines; as though in determining ionizing-potentials by elec- 

 tron-impacts one had to use electrons of certain distinct and wideh- 

 spaced energ>^-values, and could not refine the measurements by ad- 

 justing the accelerating voltage to intermediate values ad libitum. In 

 measuring ionizing-potentials by electron-impacts there is a secondar\- 

 difficulty due to radiation from struck atoms falling upon the collector; 

 here the difficulty becomes a primary one, since the primary radiation 

 itself is competent to produce this efifect. The effect is most vicious 

 with alkali-metal vapours, as they deposit themsehes over all the solid 

 surfaces of the apparatus in films excessively liable to pour out elec- 

 trons when stimulated by light or warmth; yet these are the only ele- 

 ments for which Xo lies above 2500A. 



Several experimenters have minimized the undesired effects of the 

 radiation by projecting a narrow beam of light across a jet of alkali- 

 metal vapor boiling up out of a narrow channel in the main tube. The 

 beam struck nothing except the jet and beyond it a "trap" in which 

 presumably it was totally absorbed and no part was scattered. The 

 jet passed onward, near to an electrode negatively charged to receive 

 positi\e ions. With potassium vapors, for which Xo should be 2856A, 

 R. C. Williamson found ionization commencing somewhere between 

 3100A and 2800A; H. Samuel thought that it commences between 

 2804A and 2893A; E. Lawrence concluded that it begins at 2610A.^ 

 P. D. Foote and F. L. Mohler^ detected the positive ions by their effect 

 in annulling the space-charge limitations upon the current from a hot 

 filament, after the fashion of the last-mentioned method of determining 

 ionization-potentials. Their result was somewhat unexpected; they 

 found ionization in caesium vapor at wavelengths even greater than 

 the threshold-wa\'elength. This is attributed to the same cause as 

 brings about a lowering of the apparent ionizing-potential when dense 



Thys. Rev. (2) 27, pp. 37-51 (1926); 26, pp. 197-207 (1925). 



8 E. O. Lawrence, Phil. Mag. 50, pp. 345-359 (1925); R. C. Williamson, Phxs. Rev. 

 21, pp. 107 (1923) ; H. Samuel, ZS.f. Phys. 29, pp. 209-213 ( 1924); and prior literature 

 cited in the first two. In all of the cited experiments the vapor had freshly issued 

 from condensed potassium, and may have contained a large proportion of molecular 

 aggregates, to which Lawrence attributes the diiTerence between his observed thres- 

 hold-wavelength and the calculated vo, Cf. also G. F. Rouse and G. W. Giddings, 

 Proc. Nat. Acad. Sci. 11, pp. 514-177 (1925). 



