of Emission and Absorption Lines in a Gas Spectrum, 291 



yS-line in an arc-spectrum. His estimate gives a value 

 (d\\\) = 4. 10 " 6 , and this agrees with the figures given above, 

 at least in order of magnitude. 



Much more precise, however, is the information contained 

 in Humphreys' papers in the Astrophysical Journal (see 

 especially that in 1907). He has there given very numerous 

 results for the metallic lines, and a comparison with these 

 may be helpful. 



In the first instance the phenomenon was first noticed 

 when the arc spectra of certain metals were compared with 

 the solar spectrum. The corresponding lines in the two did 

 not occupy exactly the same positions. It was concluded 

 that this was a pressure effect, and Humphreys tested the 

 result by comparing the arc spectra of the metals when 

 obtained in ordinary atmospheric air at different pressures 

 (varying between 1 and 100 atmospheres). He summarizes 

 his results verbally, and I append a few of his conclusions. 



1. The brilliance of the arc becomes much greater at 

 greater pressures. 



2. The wave-lengths of all lines (except carbon) examined 

 increased approximately proportional to the increase in the 

 pressure, up to the highest pressures used, though this in- 

 crease or shift is very different, not only for different elements, 

 but even for different lines of the same element. 



3. The amount of shift of a given line is practically inde- 

 pendent of whether it is reversed : that is, the emission and 

 absorption lines are similarly and equally affected. 



4. In general the pressure shift of the spectrum lines seems 

 to increase with the wave-length, but probably this is true 

 only of the lines of the same series. 



5. In general those lines w T hich are strongly separated by 

 a magnetic field are correspondingly largely displaced by 

 pressure and conversely, those like lines of bands, that have 

 but little if any Zeeman separation, are but slightly, if at all, 

 shifted by pressure. 



Humphreys' results, covering an enormous number of bands 

 and several substances, show a large range of values for the 

 shift. His figures give shifts ranging from (d\/\) = 1*4 x 10~ 5 

 down to (rfX/\) = 5 x 10~ 7 . This easily covers the range in 

 which our theoretical determinations fall. 



The question may, however, be asked : what have our 

 theoretical results to do with those experimentally investi- 

 gated by Humphreys ? It will be noticed that the pressure 

 in his case was that of an inert or inactive gas, the air, 

 surrounding the arc : this air was in quite an ordinary con- 

 dition, at least at a distance from the arc. Our theoretical 



U2 



