352 BELL SYSTEM TECHNICAL JOURNAL 



The molecule was invoked at once as the deus ex machina; the ioniza- 

 tion beginning beyond the proper wavelengths was supposed to be 

 ionization of molecules, with or without dissociation. So long as the 

 threshold was thought to be near 2600 or 2550, this idea was fortified by 

 the following calculation. Suppose that a photon of wavelength 

 2555A has just the energy required to split a K2 molecule into a 

 K atom, a K+ ion, and a free electron; and that a photon of 2856A 

 has just the energy required to split a K atom into a K+ ion and a free 

 electron. One easily sees that then the difference between the energies 

 of these two photons would be just the energy required to split a 

 K molecule into two neutral K atoms. The difference amounts to 

 0.5 equivalent volt. This figure agrees ^^ with independent estimates 

 of the value of the latter quantity, which is the heat of dissociation of 

 K2. The force of this agreement has just been weakened by the 

 curve of Fig. 3, showing as it does that the ionization in question 

 begins near 2700A — weakened, but not destroyed, for the ions pro- 

 duced by waves shorter than 2555 might be explained in a way which 

 the reader will easily imagine after the next two paragraphs. The 

 other alternative is, to hope that quantum mechanics will presently 

 prove that the ionization-vs-frequency curve for the potassium atom 

 ought to display both the maxima which are found. 



Return now to the curve of Fig. 2 for rubidium. On the long-wave 

 side of the limiting-frequency there is a series of peaks ; they lie at the 

 frequencies of the various members of the principal series of lines in 

 the Rb spectrum. Even more striking peaks of this sort were earlier 

 obtained with caesium by Foote, Mohler, and R. L. Chenault;^^ the 

 relevant part of one of their curves is shown as Fig. 4. 



Palpably these are phenomena of the same sort as one meets when 

 mercury is irradiated by 2537; and they signify an ionizing-process of 

 two or more stages, the first of which is excitation by the absorption of 

 a photon. There is probably no need to suppose more than two stages; 

 the energy received by the atom from the photon is always much more 

 than half of what is required to ionize. It is supposed by those who 

 have obtained the data that the process is completed by an impact of 

 fast-moving atom, one of those which by virtue of Maxwell's distribu- 

 tion have the necessary excess of energy over the relatively modest 

 mean value corresponding to the actual temperature. The relative 

 heights of the peaks would then be determined partly by the relative 

 abundance of atoms having the necessary energies, and partly by the 

 relative probabilities of the corresponding types of excitation, which 



" R. W. Ditchburn, Proc. Camb. Phil. Soc, 24, pp. 320-327 (1928). 

 ^^Phys. Rev., (2) 26, pp. 195-207 (1925); 27, pp. 37-50 (1926). 



