150 



Mr. J. Norman Lockyer. 



LIST OF MAPS. 



T. Metallic lines in stars of increasing temperature. 



II. Lines of cleveite gases in stars of increasing temperature. 

 III. Metallic lines and lines of cleveite gases in stars of increasing temperature. 



TY. Temperature ranges of different substances. 



Y. Metallic lines in cooling stars. 



YI. Lines of cleveite gases in cooling stars. 

 YII. Metallic lines and lines of cleveite gases in cooling stars. 



PLATES. 



1. Arc and spark spectra of magnesium, iron, and calcium. 



2. Photographic spectra of typical stars of increasing temperature. 



3. Photographic spectra of typical cooling stars. 



I. Introduction. 



The study of stellar spectra from the time of Rutherfurd to the 

 present shows us that only a very small number of groups is in 

 question. We seem to be in presence of an evolution in which only 

 a very few variables are in operation, and in my opinion the pheno- 

 mena suggest that the only variable of paramount importance is 

 temperature. 



In working out the classification of stellar spectra, which I com- 

 municated to the Royal Society in 1888, the course pursued was to 

 study the flutings and lines of the various elements given in the 

 existing lists and to fill up gaps in them by fresh experimental work 

 with the view of finding the necessary criteria. 



The question, however, was complicated by the discovery in stellar 

 spectra of many lines, the origin of which could not be stated. 



Some time has now elapsed since my classification was published. 

 In the meanwhile the attempts to trace the origin of the unknown 

 lines have been continued, and the discovery of terrestrial sources of 

 helium, and probably other gases, has thrown a flood of new light 

 upon stellar chemistry. 



I propose in the present paper to trace the history of the criteria 

 now at our disposal in the study of the phenomena of the hottest 

 stars and to give the results of my latest researches. In the present 

 communication, thus limited in scope, we are only concerned with 

 spectral line phenomena, and it is important that some of the main 

 points connected with line spectra in their relation to celestial bodies 

 should be referred to by way of introduction to the later results. 



The Criterion of Long and Short Lines. 



In November, 1872,* I showed that when an image of an electric 

 spark is thrown on the slit of a spectroscope the lines are seen of 



* ' Phil. Trans.,' 1873, vol. 163, Part 1, pp. 253-275. 



