438 BELL. SYSTEM TECHNICAL JOURNAL 



— iRIi is drawn as a single dash, the next le\el as two dashes, the 

 next as three, and so forth. Paschcn's data were therefore compared 

 with this prediction. 



The data and the prediction were found compatible. If arrows 

 are drawn from every "component" stationary state to every other 

 "component" stationary state, it is found that each of the lines which 

 was observed corresponds to one of the arrows (but it is necessary 

 to assume that, in some places, two or more adjacent lines are fused 

 apparently into one by reason of the insufficient resolving-power of 

 the spectroscope). Some of the arrows, however, correspond to missing 

 lines. Evidently some sort of inhibiting agency is at work; some 

 sort of a selection-principle is adumbrated. Furthermore, some and 

 perhaps all of the missing lines appear when the electric field strength 

 acting upon the radiating atoms is increased, and this, it will be 

 remembered, is the beha\ior of the missing lines in the sodium spec- 

 trum. Whether the selection-principle could ever have been inferred 

 from these data alone seems doubtful. Naturally one proceeds to 

 try out the same principle as served for the previous case. Can the 

 component stationary states of the ionized-helium atom be sorted 

 out into parallel columns, in such a manner that transitions between 

 lev'els in adjacent colunms correspond to actual, all the other trans- 

 itions to missing, lines? 



This is attempted in iIk- ni.inni-r siiown in I-"ig. 8. The result is 

 fairly satisfactory. The lines due to transitions between levels in 

 adjacent columns should by this principle be visible, and they are. 

 The lines corresponding to transitions between levels in the same 

 column, or more than one column apart, should be missing; and 

 some of them are, but also some of them undeniably can be seen. 

 To account for these unwelcome guests, it is necessary to assume 

 that some of the radiating atoms are subject to a strong electric field 

 which might, but would not be likely to, exist in the discharge. This 

 is an unconiforlable solution; but there are other numerical agree- 

 ments between the prediction and the data, which it is not expedient 

 to describe at this point, but which are good enough to excuse that 

 deficiency to some extent. Eti somme, the evidence presents no 

 insuperable objection to our arranging the component stationar\- 

 states of the ionized-helium atom in parallel columns, and declaring 

 that the only transitions which occur (except in strong electric fields) 

 are those between members of adjacent columns; and this is just 

 what we did with the sodium atom, and can in general do with every 

 other kind of atom whereof the spectrum has been interpreted. This 

 being granted, we can assert that the spectra and the stationary 



