On the Chemist?"!/ of the Hottest Stars. 



163 



Using the large coil and jars, the line has also been photographed in 

 the spectrnm of the chloride ; indeed all the lines recorded in the 

 arc by Kayser and Range have been photographed in the spark 

 spectrum of the metal. 



The work with the large jars has also resulted in the detection of 

 another line of magnesium about wave-length 4395 which does not 

 appear in the arc spectrum, and the line about 4587*4 observed by 

 Thalen also shows itself feebly. The former of these is fairly bright 

 and seems to be closely associated with 4481. 



Again there are four distinct temperature steps, namely : 



(1) The name spectrum, represented by lines at 4571*3 and b, a 

 triplet in the ultra-violet commencing with a line at 3734, and two 

 flutings, one commencing at 5210 and the other at 5006*5.* 



(2) The arc spectrum, comprising b, a line at 4352*18, and another 

 triplet in the ultra-violet commencing with 3838*4,4481 being almost 

 invisible, while 4395 and 4587 are quite invisible. 



(3) The spark spectrum, including all the arc lines, but with 

 4481 intensely bright, 439 5 fairly bright though short and 4587*4 

 rather feeble. 



(4) As 4481, 4395, and 4587*4 are much intensified in the spark 

 spectrum, we can conceive a fourth stage at a still higher temperature 

 when magnesium would be represented by these lines alone. The 

 complete spectra of magnesium under the three conditions at which, 

 observations can be made are indicated in the accompanying table 

 (p. 1.64). 



The arc and spark spectra, as photographed with the moderate 

 dispersion to which reference has already been made, are shown in 

 Plate 1, the three enhanced lines being indicated as in the case of iron 

 and calcium. 



IY. Application op the eecent Investigations to Stellae Spectea. 



In order to study the presence or absence of the enhanced lines in 

 the spectra of the stars, the Kensington series of stellar spectrographs, 

 in which the dispersion is not greatly different from that of the 

 metallic spectra, has been employed. 



Although this discussion has specially dealt with the hottest stars, 

 it is necessary to include references to the cooler ones in order to con- 

 trast the behaviour of the high temperature lines with those which 

 are characteristic of low temperatures, and further to compare the 

 appearances of the lines at different stellar temperatures. 



a Orionis is taken as a typical case of a relatively cool star which 



* Messrs. Liveing and Dewar have ascribed these fhitings to compounds of 

 magnesium -with hydrogen and oxygen respectively, but whether they are due to 

 compounds or to the metal itself is immaterial for my present purpose. 



VOL. LXI. N 



