354 



Profs. Liveing and Dewar 



[Feb. 20,. 



the tube before it was introduced into the furnace, and watching the 

 increase of light as the tube got hot, this band was at first seen bright on 

 a less bright background, it gradually faded, and then came out again 

 reversed, and remained so. ~No very high temperature was required 

 for this, but a rise of temperature had the effect of widening the band. 

 Besides this absorption, there appeared a very indefinite faint absorp- 

 tion in the red, with the centre at a wave-length of about 6,100, and a 

 dark band, with a tolerably well-defined edge on the less refrangible 

 side, at about a wave-length of 4,850, shading away towards the violet. 

 A fainter dark band was sometimes seen beyond, with a wave-length 

 of about 4,645 ; but sometimes the light seemed abruptly terminated 

 at about wave-length 4,850. It will be noticed that these absorptions 

 are not the same as those seen when potassium is heated in hydrogen, 

 nor do they correspond with known emission lines of potassium, 

 although the first, which is also the most conspicuous and regularly 

 visible of these absorptions, is very near a group of three bright lines 

 of potassium. It seemed probable that they might be due to a 

 combination of potassium with carbonic oxide. We tried the effect 

 of heating potassium in carbonic oxide in glass tubes, but, though 

 the potassium united readily with the gas, the compound did not 

 appear to volatilize at a dull red heat, and no absorption, not even that 

 which potassium gives when heated in nitrogen under similar circum- 

 stances, could be seen. We then tried induction sparks between an 

 electrode of potassium and one of platinum iu an atmosphere of 

 carbonic oxide. The usual bright lines of potassium were seen, and 

 also a bright band, identical in position with the above-mentioned band, 

 between wave-lengths about 5,700 and 5,775. This band could not be 

 seen when hydrogen was substituted for carbonic oxide. A mixture of 

 sodium carbonate and charred sugar, heated in an iron tube, gave only 

 the same absorption as sodium in hydrogen. There were also no 

 indications of any absorption due to a compound of rubidium or of 

 caesium with carbonic oxide. 



The experiments of Mallet (" Chem. Soc. J.," 1876) on the vola- 

 tility of calcium, strontium, and barium, and the reducing action of 

 aluminium on the oxides, especially in the presence of carbonate of 

 sodium, induced us to try similar mixtures in our tubes. 



A mixture of barium carbonate, aluminium filings, and lamp-black, 

 heated in a porcelain tube, gave two absorption lines in the green, 

 corresponding in position to bright lines seen when sparks are taken 

 from a solution of barium chloride, at wave-lengths 5,242 and 5,136, 

 marked a and /3 by Lecoq de Boisbauclran. These two absorptions 

 were very persistent, and were produced on several occasions. A 

 third absorption line, corresponding to line B of Boisbaudran, was 

 sometimes seen, and on one occasion, when the temperature was as 

 hi e h as could be obtained in the furnace fed with Welsh coal, and a 



