1874.] Absorption- Spectra of Potassium and Sodium. 363 



which the metal was melted in hydrogen several times on successive days 

 until no further change in the bright character of the globule could be 

 perceived. On vaporizing the metal, which had been melted down to a 

 clean portion of the tube, the bands were seen as before, and came out 

 even more clearly, the globule, after heating, exhibiting a bright metallic 

 surface. An analysis of the potassium used showed that it did not con- 

 tain more than 0'8 per cent, of sodium, although, of course, the double 

 line D was always plainly seen. 



In order to ascertain whether an alteration in the absorption-spectrum 

 of the metal takes place at a red heat, fragments of potassium were 

 placed in a red-hot iron tube, through which a rapid current of pure 

 hydrogen gas was passed, the ends of the tube being closed by glass 

 plates. The magnificent green colour of the vapour was clearly seen at 

 this temperature, on looking through the tube at a lime-light placed at the 

 other end. Owing, doubtless, to the greater thickness or increased pres- 

 sure of the vapour, the bands seen by the previous method could not be 

 resolved by the small spectroscope employed, the whole of the red being 

 absorbed, whilst a broad absorption-band in the greenish yellow was seen 

 occupying the place of the group y. 



The positions of the bands obtained by the first method were measured 

 by means of a telescope and distant scale, and the wave-lengths obtained 

 by an interpolation curve, for which well known air-lines were taken as 

 references. The following numbers give the wave-lengths of the most 

 distinct, that is, the most refrangible edge of each band. As the 

 measurements had to be made quickly, owing to the rapid darkening 

 of the glass by the action of the metallic vapour, these numbers do not lay 

 claim to very great accuracy, but fairly represent the relative positions 

 of the band, and show that they do not always occur at regular intervals, 

 although they are pretty regularly spread over the field and all are 

 shaded alike. 



Bands of potassium shaded off towards red. Wave-lengths in tenth- 

 metres : 



The bright potassium-lines in the red and violet were not seen re- 

 versed, the intensity of the lime-light being too small at both extremes to 

 render an observation possible. 



In order to ascertain whether the vapour of sodium, which, when seen 

 in thin layers, appears nearly colourless, exhibits similar absorption-bands, 



