354 BELL SYSTEM TECHNICAL JOURNAL 



convergence-frequency itself, multiplied by //, give the energy required 

 to ionize an atom from that initial state which is common to the entire 

 series. 



Thus photons having the convergence-frequency of any series are 

 just able to detach an electron from an atom in the corresponding 

 state. Consequently photons having any greater frequency have 

 energy sufficient to detach an electron, and give it some kinetic energy 

 in addition. Now we are not aware of any "quantum" limitations on 

 the amount of energy which a freed electron may receive. We thus 

 infer that light of any frequency superior to a convergence-frequency 

 will be able to ionize atoms and to be absorbed in doing so, and that 

 there will be a continuous region of absorption in the spectrum extend- 

 ing upwards from the limit of each series. For such a region I will 

 use the terms continuous band and continuum. 



Bohr drew this inference in the first of his epoch-making papers on 

 the interpretation of spectra. He was able then to point to only one 

 example; a continuum beyond the limit of the principal series of 

 sodium, observed by R. W. Wood.^^ Afterwards J. Hartmann ^^ 

 searched the spectrograms of the stars, and in those of the so-called 

 "hydrogen stars" he found a continuous band beyond the limit of 

 the Balmer series. This, be it noted, is the sign of ionization of 

 hydrogen atoms initially not in the normal, but in a certain excited 

 state. The continua beyond the principal series of the alkali metals, 

 however, are due to ionization of normal atoms. Those of sodium and 

 potassium were studied by Holtsmark; ^"^ those of caesium and rubid- 

 ium have been discerned (Harrison, I.e. infra); and the former two 

 were measured, that is to say the variation of absorption-coefficient 

 with frequency was measured, for sodium by G. R. Harrison ^^ and 

 B. Trumpy,^^ and for potassium by R. W. Ditchburn.^^ 



Obviously if the fundamental theory is correct, absorption is pro- 

 portional to ionization, and the curves representing the two as functions 

 of wavelength should coincide everywhere if scaled to coincide at any 

 one point; and measurements of either should make the other nugatory. 

 Unfortunately it is difficult to measure the absorption properly, perhaps 

 impossible to do it with anything like the precision feasible with the 

 other measurement.'-^ Harrison managed to get smooth absorption- 



'^Phil. Mag., (6) 18, pp. 530-534 (1909). 



^^Phys. ZS., 18, pp. 429-432 (1917). 



^"^ Phys. Rev., 20, pp. 88-92 (1919). 



^^Phys. Rev., (2) 24, pp. 466-477 (1924). 



13 Z5./. Phys., 47, pp. 804-813 (1928). 



2° Proc. Roy. Soc, 117, pp. 486-508 (1928). 



^1 Mohler and his colleagues state that with an amount of ionization tenfold greater 

 than that which is observed with caesium at the series-limit, a stratum of the gas at 

 230° would have to be nitie metres deep to give a 50 per cent absorption. 



