430 BELL SYSTEM TECIIMCAL JOVRSAL 



iliis iiiiii. ilii' (litTcrence between them is equal to 1 'c times the fre- 

 quency' III ilu' line whirh corresponds to the transition from one to 

 the other. 



A speclruni-line corresponding; to a transition between two station- 

 ary states is symbolized, on a diagram of statit)nary states, by an 

 arrow connecting the dashes (or whatever marks are used) which 

 symbolize the two levels. This is illustrated in Fig. 6. 



I pause at this point to remark that each of what I ha\e been 

 calling the "stationary states" is in fact usually a group of two or 

 more stationary states, often but not always exceedingly close to- 

 gether; just as many stars in the sk\- are in fact groups of stars too 

 close together for any but an excellent telescope to discriminate. 

 This will be discussed at length in a later section; at present it is 

 expedient to regard each of these groups as one stationar\' state. 



The experimental test of Bohr's method for identifying stationary 

 states consists in comparing the stationary states inferred from the 

 spectrum, according to Bohr's procedure, with the stationary states 

 derived directly by the study of electron-impacts. The agreement 

 is perfect where\'er the experiments by the latter method can be 

 carried out. By a curious fatality, this is impracticable for hydrogen 

 and ionized helium, as neither sort of atom occurs in gas quiescent 

 enough for experiments on energy-transfers from electrons to atoms. 



For about fifteen other elements, the comparison has been made 

 for two or more of the Stationary States. Every energy-value given in 

 Table II was obtained by the method of electron impacts, and con- 

 firmed b\- analyzing the spectrum of the element. 



E 3. The Classification of Stalioiiary Slates hy Vtiliziiig "Rules of 

 Selection" 



I have said that e\ery line in a spectrum, at least of those arranged 

 in series, may be represented by an arrow connecting two stationar\- 

 states. If arrows are drawn from every one of the stationary states 

 to everj' other, will every arrow correspond to a line actually observed 

 in the spectrum? Kvery line has an arrow; does everj' arrow have a 

 line? By no means; the answer is definiteh- and strongly negative. 

 If the wave lengths deduced from all the possible arrows are sought 

 in the spectrum, most o( them are found unoccupied by lines. The 

 great majority of the apparently po.ssible transitions either do not 

 occur at all, or if they do occur, the energy which is liberated is dis- 

 posed of in some other way than by radiation. There is reason for 

 believing that the atom may embrace this last alternative if it col- 



