THEORETICAL. 329 



This is not a very long column of the gas as compared with Paschen's, 

 nevertheless, judging from the behavior of the small traces of CO, in 

 air, it does seem impossible for it to emit radiation as intense as that 

 observed at 4.75 jj.. This would require a very high temperature. If 

 the shifting of the COo band toward the long wave-lengths continues 

 with rise in temperature for the region beyond 4.4//., just as Paschen 

 (loc. cit.) found for the region preceding 4.4/*, then the 4.75 ju. band 

 would indicate a temperature of some 6,500° to 7,000° (found by 

 extrapolating from Paschen's values). This is close to Stark's (loc. 

 cit.) "electrical temperature" of 6,000° for the cathode glow. Return- 

 ing to the strong emission lines just at the end of the red, if we consider 

 the maximum of the envelope (the curve) drawn through the highest 

 points on these emission lines, which maximum lies just beyond the red, 

 then, from the "displacement law," A^a^T=const, the thermal tempera- 

 ture appears to be about 4,000° abs. 



From this line of reasoning it would appear that we can consider the 

 4.75 jj. band and the bands at the end of the red to be due to a high 

 thermal condition in the vacuum-tube, without having recourse to an 

 "electrical temperature." 



The continuous spectrum of alcohol vapor would indicate a higher 

 temperature than that found by bolometric measurements. But even 

 here the evidence is contradictory when compared with the emission of 

 water vapor which showed no emission spectrum at 2.8 /i, where the 

 Bunsen flame has emission lines. 



Evidently further investigation is needed to elucidate this subject — 

 and such an investigation is in progress. 



Since submitting this paper for publication a similar investigation of 

 CO, and N, by Drew (Phys. Rev., 21, p. 122, 1905) has appeared. He 

 studied the emission band at 4.69/* (the 4-75/^ band in the present 

 work) but did not succeed in eliminating it from N, which was made 

 from sodium nitrate, as in the present work. Unfortunately he did not 

 examine CO, and he reached a less definite conclusion as to the source 

 of this line. He drew straight lines through the observed points, which 

 gave an isosceles triangle, and from this he thinks that an observed 

 shift of 0.02 /x for a change in pressure of the gas is real. His deflec- 

 tions were much smaller than in the present work, and the observed 

 points do not always lie close enough upon the lines to convince one 

 of the reality of the shift. Such a shift of the maximum from 4.67 fi for 

 a pressure of 3 mm. to 4.695 /x for a pressure of 0.6 mm. is to be ex- 

 pected if CO2 dissociates into CO with decrease in pressure. In the 

 present work the deflections were very much larger, but even here the 

 variations in the readings are too great to be certain of a shift of the 

 order of 0.02 /*. 



