136 On the Reversal of the Lines of Metallic Vapours. [Feb. 28, 



gible than that double line. In the next stage, the channelled-space 

 spectrum comes out in the dark space between the green and blue, 

 and finally, in the red. Gradually the light extends, the channels 

 disappear, the D line absorption narrows, but still the dark line 

 in the green is plainly discernible. Lastly, there is only D lines 

 absorption. So far as we can judge, the blue and the streak of green 

 light at first observed are due to luminosity of the vapour itself, where 

 it is somewhat cooled, the later stages being mixed phenomena of 

 absorption and emission. 



As the absorption line, with wave-length about 5,510, which we may 

 call for shortness the second absorption line of sodium, has not been 

 distinctly recorded by other observers, we have endeavoured to trace 

 it under somewhat different conditions from that of the vapour vola- 

 tilized in white hot iron tubes. This absorption-line is easily seen 

 when a gas flame is observed through a horizontal glass tube, about 

 three inches long, containing sodium volatilized in the middle of the 

 tube by the heat of a Bunsen's burner, and equally well whether the 

 tube contains hydrogen or nitrogen besides sodium. We have also 

 observed the same absorption line when a piece of commercial mag- 

 nesium ribbon (which always contains sodium) is ignited in a hori- 

 zontal position, so that the metal melts and produces an elongated 

 flame. It is of some interest to note that absorption lines of about this 

 wave-length, in the solar spectrum, are given by Kirchoff and Angstrom 

 not corresponding with emission lines of known elementary bodies. 



When potassium vapour is observed, whether in the iron tube or in 

 a glass tube, an absorption line is seen, with a wave-length of about 

 5,730, which is more refrangible than the yellow double emission line 

 of potassium, and does not correspond to any known bright line of 

 that metal. 



We reserve, for a future communication, the discussion of the identity 

 or non-identity of these absorption lines with lines in the solar spectrum 

 and the inferences which may be drawn from such determination. 



The method of observation we have described may be used to observe 

 emission spectra as well as absorption spectra, for if the closed end of 

 the tube be placed against the bars of the furnace so as to be relatively 

 cooler than the middle of the tube, the light emitted by the vapours in 

 the hottest part is more intense than that emitted .by the bottom of 

 the tube. This succeeds admirably with sodium, but we have not 

 specially observed it with other vapours. 



