350 BELL SYSTEM TECHNICAL JOURNAL 



at the limiting frequency in each case. Earlier E. M. Little ' had got 

 a value two orders of magnitude lower for caesium; this difference has 

 not been reconciled. These values will later be compared with those 

 to which the theories lead. 



The upturn in the curve of Fig. 1 on the shortwave side of 2600A 

 may serve ^ as an introduction to the case of potassium. Adjourning 

 therefore the discussion of the righthand part of the curve of Fig. 2, 

 I take up next this strange and singular case. 



The first who plotted an ionization-vs-wavelength curve for potas- 

 sium was E. O. Lawrence.^ The vapor-pressure of potassium being 

 low, he so designed his tube that the beam of light passed across the 

 vertically-rising jet of gas distilling from a pool of highly-heated metal. 

 This expedient was used by all the other physicists who worked upon 

 potassium, and was at one time held responsible for the curious results, 

 until finally Mohler and Boeckner confirmed the previous data by 

 measurements on stagnant vapor. The ionization-current was col- 

 lected by electrodes placed on either side of the jet and away from 

 the light; so the method is fit to give the relative ionizing-powers of 

 light of various wavelengths, though not an absolute measurement, 

 the density in the jet being unknown. Lawrence's monochromator 

 provided beams of light extending over some 80A of the spectrum. 



Few data can have been more unexpected, indeed more positively 

 unwelcome, than those which he obtained; for what they intimated 

 was, that ionization begins, or at least the sharp increase of ionization 

 occurs, at a wavelength definitely too small. It seems as though a photon 

 could not ionize a potassium atom without having definitely more than 

 the necessary energy; a conclusion which would be in disaccord with 

 fundamental theory, and with the (subsequent) experiments upon 

 rubidium and casium. 



New experiments upon potassium gave comfort to the theory, but 

 also demonstrated the anomaly which Lawrence had discovered.'" The 



7 Phys. Rev., (2) 30, pp. 109-118, pp. 963-964 (1927). 



* However it does not appear in the corresponding curve obtained by Lawrence 

 and Ediefsen. 



^Phil. Mag., 50, pp. 345-359 (1925). There had been four precursors: S. H. 

 Anderson, L. A. Gilbieath, R. C. Williamson, R. Samuel (for the references, see 

 Hughes, I.e.). The earliest two reported ionization at wavelengths where it now 

 seems unlikely that true ionization of the vapor would have been perceptible; the 

 others used chemical filters and so were unable to plot a curve, but seem to have 

 observed the weak ionization produced between 2800 and 3100A. 



" Such a proof would relieve us from one of the greater difficulties of the " molecule " 

 hypothesis — the necessity of assuming that ionization of a K2 molecule by light is an 

 event thousands of times as probable as that of a K atom, for in potassium vapor 

 under the actual conditions free atoms are believed to be a thousand times more abun- 

 dant than molecules, and yet the ions which we are ascribing to the latter are much 

 more plentiful. R. W. Ditchburn and F. L. Arnot {Proc. Roy. Soc. 123, pp. 516-536 

 (1929)) found nothing but K+ ions in the ionized vapor, thus disposing of the notion 

 that the process might consist in the detachment of an electron from a thenceforward 

 stable K2 particle. 



