November io, 1898] 



NA rURE 



ZZ 



In dealing with my subject, I shall first refer to the 

 work which has been done in more recent years with 

 regard to this chemical conditioning of the atmospheres 

 of stars, and afterwards very briefly show how this work 

 carries us into still other new and wider fields of thought. 

 The first important matter which lies on the surface 

 of such a general inquiry as this is that if we deal with 

 the chemical elements as judged by the lines in their 

 spectra, we know for certain of the existence of oxygen, 

 of nitrogen, of argon, representing one class of gases, in 

 no celestial body whatever ; whereas, representing other 

 gases, we have a tremendous demonstration of the exist- 

 ence of all the known lines of hydrogen and helium. 



We see then that the celestial sorting out of gases is 

 quite different from the terrestrial one. 



Taking the substances classed by the chemist as non- 

 metals, we find carbon and silicium — I prefer, on account 

 of its stellar behaviour, to call it silicium, though it is 

 old-fashioned — present in celestial phenomena ; we have 

 evidence of this in the fact that we have a considerable 

 development of carbon in some stars and an indication 

 of silicium in others. But these are the only non-metals 

 observed. Now with regard to the metallic substances 

 which we find, we deal chiefly with calcium, strontium, 

 iron and magnesium ; others are not absolutely absent, 

 but their percentage quantity is so small that they are 

 negligible in a general statement. 



Now do these chemical elements exist indiscriminately 

 in all the celestial bodies, so that practically, from a 

 chemical point of view, the bodies appear to us of similar 

 chemical constitution.' No, it is not so. 



From the spectra of those stars which resemble the 

 sun, in that they consist of an interior nucleus surrounded 

 by an atmosphere which absorbs the light of the nucleus, 

 and which therefore we study by means of this absorp- 

 tion ; it is to be gathered that the atmospheres of some 

 stars are chiefly gaseous, /.(•. consisting of elements we 

 recognise as gases here, of others chiefly metallic, of 

 others again mainly composed of carbon or compounds 

 of carbon. 



Here then we have spectroscopically revealed the fact 

 that there is considerable variation in the chemical 

 constituents which build up the stellar atmospheres. 



This, though a general, is still an isolated statement. 

 Can we connect it with another? Oneof the laws fonnu- 

 lated by Kirchhoff in the infancy of spectroscopic inquiry 

 has to do with the kind of radiation given out by bodies 

 at different temperatures. .\ poker placed in a fire first 

 becomes red, and as it gets hotter, a'///A', hot. Examined 

 in a spectroscope we find that the red condition comes 

 from the absciuf of blue light ; that the white condition 

 conies from the gradual addition of blue as the tempera- 

 ture increases. 



The law affirms that the hotter a mass of matter is the 

 further its spectrum extends into the ultra-violet. 



Hence the hotter a star is, the further does its com- 

 plete or contimwus spectrum lengthen out towards the 

 ultra-violet, and the less is it absorbed by cooler vapours 

 in its atmosphere. 



Now to deal with three of the main groups of stars, we 

 find the following very general result : — 



Gaseous stars 

 Metallic stars 

 Carbon stars 



Longest spectrum. 

 Medium spectrum. 

 Shortest spectrum. 



We have now associated two different series of pheno- 

 mena, and we are enabled to make the following 

 statement : — 



Gaseous stars 

 Metallic stars 

 Carbon stars 



Highest temperature. 

 Medium temperature. 

 Lowest temperature. 



Hence the differences in apparent chemical constitu- 

 tions are associated with differences of temperature. 

 NO. 151 5, VOL. 59] 



Can we associate with the two to which I have already 

 called attention still a third class of facts ? 



Laboratory work enables us to do this. When I 

 began my inquiries the idea was, one gas or vapour one 

 spectrum. We now know that this is not true ; the 

 systems of bright lines given out by radiating substances 

 change with the temperature. 



We can get the spectrum of a well-known compound 

 substance — say carbonic oxide ; it is one special to the 

 compound ; we increase the temperature so as to break 

 up the compound, and we then get the spectra of its 

 constituents, carbon and o.xygen. 



But the important thing in the present connection is 

 that the spectra of the chemical elements behave exactly 

 in the same way as the spectra of known compounds do 

 when we emplov temperatures far higher than those 

 which break up the compounds ; and indeed in some 

 cases the changes are more marked. For brevity I will 

 take for purposes of illustration three substances, and 

 deal with one increase of temperature only, a consider- 

 able one and obtainable by rendering a substance incan- 

 descent, first by a direct current of electricity, as happens 

 in the so-called " arc lamps " employed in electric lighting, 

 and next by the employment of a powerful induction coil 

 and battery of leyden jars. In laboratory parlance we 

 pass thus froin the arc to the jar-spark. In the case of 

 magnesium, iron and calcium, the changes observed on 

 passing from the temperature of the arc to that of the 

 spark have been minutely observed. In each, new lines 

 are added or old ones are intensified at the higher 

 temperature. Such lines have been termed enhanced lines. 



These enhanced lines are not seen alone : outside the 

 region of high temperature in which they are produced, 

 the cooling vapours give us the cool lines. Still we can 

 conceive the enhanced lines to be seen alone at the 

 highest temperature in a space sufficiently shielded from 

 the action of all lower temperatures, but such a shielding 

 is beyond our laboratory expedients. 



In watching the appearance of these special enhanced 

 lines in stellar spectra we have a third series of pheno- 

 mena available, and we find that the results are absolutely 

 in harmony with what has gone before. Thus 



Gaseous stars ... Highest temperature... -' S'/°"e J?^"""; ,. ^""^ 

 ^ r y faint enhanced lines. 



j' Feeble helium and 



Metallic stars... Medium temperature -' strong enhanced lines. 

 •^ I -No helium ana strong 



1. arc lines. 

 Carbon stars .. Lowest temperature... Faint arc lines. 



It is clear now, not only that the spectral changes \n 

 stars are associated with, or produced by, changes of 

 temperature, but that the study of the enhanced spark 

 and the arc lines lands us in the possibility of a rigorous 

 stellar thermometry, such lines being more easy to 

 observe than the relative lengths of spectrum. 



Accepting this, we can take a long stride forward and, 

 by carefully studying the chemical revelations of the 

 spectrum, classify the stars along a line of temperature. 

 But which line ? Were all the stars, in popular phrase- 

 ology, created hot? If so, we should simply deal with 

 the running down of temperature, and because all the 

 hottest stars are chemically alike, all cooler stars would 

 be alike. But there are two very distinct groups of coolest 

 stars ; and since there are two different kinds of coolest 

 stars, and only one kind of hottest star, it can not be 

 merely a question cither of a running up or a running 

 down of temperature. 



Many years of very detailed inquiry have convinced 

 me that all stars save the hottest must be sorted out into 

 two series— those getting hotter and those, like our sun, 

 getting cooler, and that the hottest stage in the history 

 of a star is reached near the middle of its life. 



The method of mquiry adopted has been to compare 

 large-scale photographs of the spectra of the differ 



