2 6o 



NA TURE 



\yan. 12, 1888 



A 



DUNER ON STARS WITH SPECTRA OF 

 CLASS III. 1 



IL 



SERIES of observations such as ours ought to add at least a 

 little to our knowledge of the development by which the spec- 

 tra of stars pass from the sec md class to one of the two sections 

 of the third, especially if these observations are combined with 

 those made of the stars of the two first classes generally, and of 

 our sun in particular ; we might even draw conclusions as to the 

 successive development of stars after they have already reached 

 this class. He who sees trees in a forest in different stages of de- 

 velopment, some old, some young, some decaying, can at once 

 form an idea of the different stages undergone by each : it is just 

 the same with the observer of the different classes of stellar 

 spectra. 



The spectra of the first cla5S are characterized by the almost 

 total absence of all metallic lines excepting those of hydrogen. 

 In spite of that, we cannot doubt for a moment the presence of 

 metallic gases in their atmospheres, for even in the spectrum of 

 Vega we can faintly distinguish the principal rays of sodium, 

 magnesium, and iron. But these gases are probably at such a high 

 temperature that their power of absorption is very slight. But 

 as the star cools and the spectrum approaches the second class, 

 the metallic lines become stronger and more numerous, whilst, 

 strange to say, the lines of hydrogen diminish. Thus the spec- 

 trum becomes more and more like that of our sun in its actual 

 state, and at length, as the metallic lines increase, it resembles 

 that of Arcturus. 



Up to this stage of development it is unnecessary to consider 

 the two divisions of the third class separately, but after this it 

 becomes indispensable. 



In those spectra which at length become Ill.a, the change 

 seems to operate as follows. On account, probably, of the 

 progressive cooling, the metallic lines, especially those of iron, 

 magnesium, calcium, and sodium, become larger, and, besides 

 these, numerous weak narrow lines are seen grouped together, 

 generally in the neighbourhood of the stronger lines. At this 

 stage it is often difficult, if not impossible, to decide, with spec- 

 troscopes of small dispersion, whether one sees broad lines or 

 real bands (or fiatin^s). This happens in the spectrum of 

 Aldebaran. The faint lines go on accumulating, until they can- 

 not be separated from one another and occupy broader spaces, 

 and now the spectruai is easily seen to belong to Class II La. 

 At first the bands in the red and orange are the only ones dis- 

 tinctly visible ; but later the bands in the green-blue and in 

 the blue become very strong and broad. 



While the development of the stars 111. a was very well known 

 before my researches, former observers have known no star 

 with a spectrum intermediate between II. a and 111.3. Thus, 



ji , . ' Continued from p. 23'. 



M. Pechiile declares the hypothesis of the co-ordination of the 

 III. « and III. ^ classes to be inadmissible. On the other hand, 

 he seems disposed to think that the spectra III./^ represent a 

 phase, perhaps the last before i*s total extinction, in the deve- 

 lopment of each star, and that the passage from type I II. « to 

 IW.h takes place suddenly or by a catastrophe, during which 

 the bright lines appear ("Expedition Danoise," pp. 22-25). 

 M. Pechiile seems, however, to consider this hypothesis doubt- 

 ful, and at length declares that the physical role of the stars 

 III. 3 is still quite a mystery. 



A very simple explanation clears up at least part of this mys- 

 tery. If the hypothesis which I, in full agreement with M. 

 Vogel, have suggested be correct, the stars intermediate 

 between the seconcl and third classes must necessarily be com- 

 paratively rare, considering that this is only a transitory phase 

 of their existence. The general spectroscopic observations of 

 M. Vogel affirm this fact, for amongst the numerous stars 

 examined by him there are only forty-eight whose spectra are 

 denoted by 11. a I ! ! ll.a ! ! or II. a ! But as the lines must 

 be very distinctly visible in the spectra of the stars which are on 

 the point of passing from the second class to Class III. a, we are 

 obliged to acknowledge that almost all the stars of this category 

 within the zone examined by M. Vogel are among these forty- 

 eight objects. At first sight one might be disposed to seek these 

 stars amon'3' those whose spectra are designated by M. Vogel 

 by ll.a {Ul.a), Il.a? lll.a, and IILrt! (Il.a); but a clo^-r 

 examination shows that although it is not impossible that these 

 spectra may be among these objects, they must be so rare that 

 that is of no essential consequence as regards the question which 

 occupies us. 



Amongst these stars there are none which attain the magni- 

 tude 4'5, and only fourteen which surpass the magnitude 6*4. 

 All the others are faint objects, and the ambiguous symbols show 

 the difficulty M. Vogel found in recognizing with certainty the 

 details in the spectra, and not that he could not decide with 

 certainty to which of the two contiguous classes a spectrum of 

 which he could easily perceive the details belongs. The correct- 

 ness of this supposition is, however, proved by the circumstance 

 that certain spectra are designated by III.«(III./'),or III.«? lll.li. 

 And none will believe that M. Vogel meant to imply that these 

 spectra were in the act of passing f^i-om one section of the thir 1 

 class to the other. Besides, one of these stars is R Serpentis, 

 whose spectrum when the star is at the maximum is one of the 

 most strongly marked of lll.a, according to M. Vogel's earlier 

 researches, and according to mine. But in his general spectro- 

 scopic review M. Vogel examined it when its magnitude was 

 only 9"0, and therefore it was easy to doubt, on account of 

 the excessive width of the bands, whether the spectrum m'ght 

 not be lll.b instead of III. a. 



Consequently, although I think I am right in admitting that 

 most of these stars belong to the pure type ll.a or Ill.«, 

 I will nevertheless suppose that a third of them really 

 have spectra intermediate between II. « and Ill.rt. Their 

 number in M. Vogel's catalogue is 1 20, and the third is 40, so 

 we should have therefore between the Pole and -25' declina- 

 tion 160 spectra intermediate between II. rt and lll.a. I found also 

 by special observations that among the spectra designated by 

 Il.rt! ! I ! Il.rt : ! and IL« ! a fourth part really belong to the 

 intermediate type. Thus there would be in all 200 such spectra, 

 a number evidently much too great. Then, the spectra III./; being 

 about fifty times rarer, we should have at most four spectra 

 intermediate between II. rt! and lll.b, and if only stars of a higher 

 magnitude than 6'o are reckoned, there would scarcely be one. 



But, if we con-ider the differences between the spectra Ill.rt 

 and lll.b, we shall find that in reahty we can scarcely expect to 

 find any spectrum intermediate between ll.a and II 1. 1^. As we 

 have seen above, the spectra lll.a are formed by the exaggeration 

 of the essential characteristics of the spectra ll.a. There must 

 then be a phase, especially if the star is not very bright, in which 

 one cannot decide to which of the two classes the spectrum 

 belongs. Thus in the spectra lll.b there are undoulotediy well- 

 marked Fraunho'er lines — for instance, D, and the narrow band 8, 

 which is probably nothing but the collection of strong lines in 

 the neighbourhood of E, and the very narrow band 5 (A. — 576/u) 

 which is almost like a broad line ; but all these details are only 

 secondary. The essential characteristics are the three nebulous, 

 very broad flutings, which owe their origin to some carbon com- 

 pound. If these hands are visible, the spectrum is called lll.b ; 

 if they are not, it is called ll.a. The only forms intermediate 

 between the spect'-a of the type of Aldebaran and the normal 



