CONTEMPORARY ADVANCES IN PHYSICS 347 



cannot suppose that the atom after its second gulp of energy picks up 

 the remaining 0.8 volt in a collision with a fast-moving ordinary atom, 

 for at normal temperatures such fast-moving atoms would be exces- 

 sively rare. Houtermans suggests that when a 2^i'o and a 2^Pi atom 

 collide with one another, they unite to form an ionized molecule Hg2+. 

 This is far from being the only case in which a molecule is invoked 

 as the deus ex machina to help out with an otherwise untenable theory. 



We turn now to the alkali metals, or rather to the three heavier 

 among them, caesium, rubidium, and potassium. With these it is 

 more nearly possible to get a full view of the situation. The phenom- 

 ena are not confined to spectrum ranges in or beyond the remotest 

 attainable fringes of the ultra violet. Indeed, in these four cases, 

 even the wavelength where ionization by single impact should begin 

 is well within reach, being in the nearer ultra violet; 241 2A for Na, 

 2856 for K, 2968 for Rb and 3183 for Cs. Ionization currents are 

 provoked by light at even greater wavelengths ; this resembles the case 

 of mercury irradiated by 2537, and is equally perplexing, indeed more so. 

 They are however much greater, near or beyond the limiting wave- 

 length for one-stage ionization; and there, we seem to be witnessing 

 the simplest process of all. With caesium, rubidium and sodium, the 

 data in this range conform to simple theories in a gratifying way. I 

 will consider these first, and then the most mysterious case of all, that 

 of potassium. 



The vapor pressures of the alkali metals increase with atomic num- 

 ber, and for rubidium and caesium are great enough to permit the 

 methods employed with the gases mentioned above: which is to say, 

 that stationary vapor of known density may be illuminated by light 

 of known intensity, and the amount of ionization be measured absolutely 

 by drawing off all the ions. This I denote as the "absolute" method. 

 There is another, the "method of space-charge annulment." The 

 tube containing the vapor contains also a hot-filament cathode and 

 some form of anode, and the filament is kept so hot, the P.D. between 

 it and the anode kept so low, that the electron-borne current between 

 cathode and anode is limited by its own space-charge. W^hen 

 positive ions are formed in the vapor, as in these experiments they are 

 by light, a fraction of the negative space-charge is annulled, and the 

 current increases. The change in the current is a measure of the num- 

 ber of positive ions formed. Nothing of the sort results if light falls 

 on solid objects in the tube and ejects electrons, an insensitiveness 

 which is a great advantage of the method. For positive ions it is a 

 very sensitive method; one finds such statements as "each positive ion 

 formed causes a million extra electrons to flow from cathode to anode," 



