On the Chemistry of the Hottest Stars. 



161 



temperature of the spark, the observations of Messrs. Kayser and 

 Runge have been limited to the arc spectrum, while my own investi- 

 gations have included all conditions of temperature available in 

 laboratory experiments. 



As I showed in 1876,* the most characteristic low temperature 

 line is that at \ 4226*9, while the H and K lines are pre-eminent at 

 high temperatues. The new work with the spark from the large 

 intensity coil and large jars has shown that all the lines recorded by 

 Kayser and Runge in the arc spectrum appear also in the spark spec- 

 trum, but with the exception of H and K, and two lines at wave- 

 lengths 3706*18 and 3737*08, which do not appear to have been 

 previously recorded in the spark, they appear with reduced relative 

 intensities. 



These two ultra-violet lines are enormously enhanced in the spark, 

 as will be seen on reference to the comparison photograph of the arc 

 and spark spectra given in Plate 1. 



As in the case of iron, four temperature steps can be recognised. 



(1) The flame spectrum, in which the blue line 4226*9 is predomi- 

 nant, H and K, and a few other lines being very feeble. 



(2) The arc spectrum, in which the H and K lines are of about 

 the same brightness as the blue line, while other feebler lines also 

 appear. 



(3) The spark spectrum, in which nearly all the lines of the arc 

 spectrum are seen, but with reduced intensities, except in the case of 

 H and K, which remain very bright, and two lines at 3706*18 and 

 3737*08, which are also very bright. 



(4) A spectrum consisting of the two lines at 3706*18 and 3737*08 

 and the H and K lines, corresponding to a temperature higher than 

 the average temperature t)f the spark, as before explained. 



The complete spectra actually recorded are shown in the following 

 table (p. 162). 



Magnesium. 



Among other substances investigated in my earlier work was 

 magnesium. 



I showed in 1879| that in the flame spectrum the two less re- 

 frangible members of the b group were seen associated with a less 

 refrangible line at 5210, making a triplet with them, while a line in 

 the blue at wave-length 4571*3, and a series, of flutings were also 

 seen ; on passing the spark, the blue line of the flame disappears, as 

 well as the flame companion to b, while two new blue lines make 

 their appearance at wave-lengths 4481*3 and 4703*3. 



Among the spark lines is one at \ 4481, of which I had previously 



* ' Eoy. Soc. Proc.,' vol. 24, p. 352. 

 f Ibid., vol. 30, p. 22. 



