October 20, 1904] 



NA TURE 



611 



lUE CLASSIFICATION OF STARS ACCOKD- 

 I.\G TO THEIR TEMPERATURE AND 

 CHEMIST R Y. 



I. 



A LTHOUGH the observations made by Fraunhofer in 

 1814 first indicated that the spectra of stars were not 

 all of the same character, it was the more systematic observ- 

 ations of Rutherfurd and Secchi fifty years later which 

 revealed the fact that the different varieties of stellar spectra 

 were, generally speaking, associated with stars of different 

 colours. The stars with fluted spectra, for instance, were 

 generally found to be red ; those resembling the sun in 

 having abundant metallic lines were yellowish ; while those 

 in which the chief absorption was due to hydrogen were 

 white. Closely following these observations came Zijllner's 

 suggestion that the spectra mieht indicate the relative ages 

 of the stars, and that the yellow and red stars were older 

 and cooler than the white ones, thus giving birth to the now 

 generally accepted view that the different kinds of stellar 

 spectra represent different temperature stages in the evolu- 

 tion of more or less similar masses of matter. iVIore direct 

 evidence as to temperature differences was brought forward 

 shortiv after by Angstrom, who directed attention to the 

 probability that the flutings characteristic of the red stars 

 originated in chemical compounds, and pointed out that the 

 occurrence of flutings in such a star as Betelgeuse might 

 be taken as an indication that the temperature of the star 

 was sufficiently reduced to permit the formation of chemical 

 combinations. Subsequent researches have shown that all 

 flutings do not proceed from compounds, but the fact re- 

 mains that in laboratory experiments flutings are only pro- 

 duced by relatively cool vapours and gases, and their 

 presence in the spectrum of a star may therefore be still 

 accepted as evidence of greatly reduced temperature. The 

 broad distinction between the spectra of cool and hot stars 

 was thus early recognised, but it remained to establish the 

 sequence of temperature in the stars characterised by line 

 spectra. 



It was next pointed out by Sir Norman Lockyer in 1873 ' 

 that the spectrum of the sun was intermediate between the 

 more coniple.\ fluted spectrum of the red stars and the 

 simpler line spectrum of the white ones, and further that the 

 great development of the blue end of the spectrum in the 

 white stars, as contrasted with stars like the sun, afforded 

 strong presumptive evidence that the white stars were the 

 hotter. E.xperiments had, in fact, shown that the con- 

 tinuous absorption exerted by certain gases was restricted 

 to the most refrangible part of the spectrum when the 

 density was low, and advanced gradually into the visible 

 spectrum as the pressure was increased. Utilising this 

 criterion, it thus appeared that the hotter a star the simpler 

 was its spectrum, and it was pointed out also that the 

 metallic elements seemed to make their appearance in the 

 order of their atomic weights. As a working hypothesis, 

 founded primarily on results obtained in solar inquiries, it 

 was suggested that in the atmospheres of the sun and stars 

 various degrees of dissociation were at work, so that in 

 some cases the atoms which compose what at terrestrial 

 temperatures we distinguish as metals, metalloids, and 

 compounds, were prevented from coming together. Hence 

 *' the so-called elements not present in the reversing layer 

 of a star will be in course of formation in the coronal atmo- 

 sphere, and in course of destruction as their vapour densities 

 carrv them down ; and their absorptions will not only be 

 smail in consequence of the reduced pressure in that region, 

 but what absorption there is will probably be limited wholly 

 or in great part to the invisible violet end of the spectrum." 



Secchi 's classification was, of course, made quite in- 

 dependently of such considerations as to temperature ; but 

 being based to a great extent on the colours of the stars 

 associated with the different spectra, the numerical sequence 

 of his four well known types is more or less in accordance 

 with the probable temperature gradation. 



Vogel ' was the first to propose a classification professedly 

 depending upon the supposition that the spectrum is in- 

 dicative of the phase of development which a star has 

 reached, and making use of the condition of the blue end 



« 1 Pi,7. Tra 

 p. 1357 (1873). 



'ol. clxiv., p. 49a (1874), and Conptes rendus, vol. Ix 

 2 Ast. Nack., vol. Ixxxiv. (1874), p. 113. 



NO. 1825, VOL. 70] 



of the spectrum as a guide to the temperature conditions. 

 In stars of his class i. the more refrangible portions of the 

 spectrum are of conspicuous intensity, in class ii. the blue 

 and violet are weaker, while in class iii., which includes 

 Secchi's third and fourth types, this part of the spectrum 

 is described as being strikingly feeble. This is, indeed, the 

 principal feature which is common to the several sub- 

 divisions of each of the three classes, and, apart from such 

 possible resemblance, it is difficult to understand, for 

 example, how stars so widely different as Arcturus and the 

 bright line stars of the Wolf-Rayet group could have been 

 brought together in the same class. Thus, although the 

 idea underlying the classification was that of decreasing 

 temperature in passing from the first to the third class, 

 there was no adequate attempt to define the successive 

 positions of the various subclasses on the descending scale 

 of temperature. 



Another idea was put forward in 1887 by Sir Norman 

 Lockver in connection with the meteoritic hypothesis.' 

 Hitherto the generally accepted view as to stellar evolution 

 had started with the assumption that all the stars were 

 intensely hot to begin with, and that all further develop- 

 ment was brought about by reduction of temperature ; but 

 it was objected that all bodies in the universe cannot^ be 

 finished suns in the ordinary sense, and that the old view 

 took no account of the processes of manufacture from nebula 

 to sun. It was then suggested that the progress of stellar 

 development was from comparatively cool nebula:, through 

 uncondensed " stars " of rising temperature, to the hottest 

 stars, with a subsequent decline, through stars like the 

 sun, to planetary conditions. On this modified basis a new 

 classification was proposed in which seven groups were 

 found sufficient to include the data depending on the visual 

 observations, which were then practically all that were 

 available. Some such arrangement of the stars in_ two 

 series is, in fact, demanded by thermodynamical principles, 

 since a mass of gas condensing under the influence of 

 gravitation must continue to rise in temperature so long as 

 it remains in a condition approaching that of a perfect gas, 

 and Prof. Darwin has shown that a condensing swarm of 

 meteorites would behave in a similar manner. 



The magnificent success which soon after attended Prof. 

 Pickering's photographic application of Fraunhofer's 

 method of studying stellar spectra by means of an objective 

 prism, and the subsequent use of the same form of instru- 

 ment by Sir Norman Lockyer and others, provided data 

 for a far more searching inquiry into the processes of stellar 

 development. Conclusions as to the relative temperatures 

 of the stars could now be more certainly drawn from the 

 extension of their spectra towards the ultra-violet, as shown 

 by the photographs, and the chemical changes accompany- 

 ing the variation of temperature from star to star could 

 be much more accurately observed. 



In a discussion of the photographic spectra of 171 of 

 the brighter stars. Sir Norman Lockyer = again found it 

 necessary to arrange the stars in an ascending and a de- 

 scending temperature series, as was previously the case 

 when dealing with the visual observations, and the general 

 sequence of events demanded by the meteoritic hypothesis 

 was therefore so far confirmed. The classification into 

 seven groups was still retained, but various subgroups were 

 introduced in order to include the finer shades of difference 

 revealed by the photographs. 



At this stage of the inquiry many of the stellar lines, 

 especiallv in the case of the hotter stars, had not been 

 identified with terrestrial spectra, and further progress 

 resulted rather from laboratory than from observatory work. 

 Sir William Ramsay's discovery of terrestrial helium per- 

 mitted a complete study of the spectrum of that element, 

 and provided a most satisfactory explanation of many of 

 the previously unknown lines appearing in the spectra of 

 some of the white stars, and other lines usually associated 

 in the stars with those of helium were subsequently traced 

 to oxygen, nitrogen, carbon, and silicium. 



But there was still another great class of outstanding 

 lines, occurring in such stars as Sirius and a Cygni, for 

 which chemical origins could not be certainly assigned on 

 current principles. Continuing his researches, dating from 



p. 117. 

 . A 1893), pp. 675-726- 



