Jan. 12, 1888] 



NATURE 



261 



type 1 11./^ are those in which the bandi are more or less faint, 

 or^eveu scarcely perceptible. In fact, I have proved not only 

 that there are spectra in which the principal bands, and especially 

 band 6, are weak on account of the briglitiiess of the stars, but I 

 have found a spectrum which is scarcely a spectrum \\\.l> yet, 

 but in which the characteristics of this class are undoubtedly 

 l)resent. 



This star is DM. + 38' 3957 — 541 Birm. In its spectrum 



rianche, Fig. 6) I have seen a rather broad and well-marked 



ii;md, whose approximate wave-length is SlQ^u, and the spectrum 



terminates abruiHly at 475ju. These wave-lengths are, within the 



Hmits of probable errors, the same a-; those of the less refrangible 



Is of bands 9 and 10 in the spectra III./'. Once I thought 



perceived a very faint trace of light beyond 475/*, and in the 



iiest atmospheric conditions I caught a glimpse of faint traces of 



the bands 4 and 6. Unfortunately the star is only of the 



eighth magnitude, so that only few details of its spectrum 



can be seen with a telescope like ours. Nevertheless, wiiat I 



(lid see seems to me of some importance in explaining the 



development of a spectrum W.a into \\\.b. 



If this spectrum be compared with those of other stars of the 

 line or even of 'a lower magnitude, such a- 145 Schj. DM. 

 ; 34° 56, DM. -f 36' 3168, it is at once seen that in the former 

 the principal bands are still in a very low stage of development, 

 and if the bands had only been a little paler nothing unusual 

 would have been seen in the s))ectrum under ordinary atmo- 

 spheric conditions. The aspect of this star seems to prove what 

 I said above, that there is, properly speaking, no intermediate 

 state between the spectra W.a and III./', but that the passage 

 from one to the other is already accomplished before the first 

 traces have been perceived. 



lUit there is still one more circumstance deserving of attention, 

 which may perhaps lead to the knowledge of other spectra which 

 are still nearer to the critical point ; that is, the very strong ab- 

 sorption of the more refrangible rays, which makes the whole 

 spectrum very short, and gives to the star itself its bright orange 

 ijolour. We know that there are many stars of a deep colour 

 id with short spectra, but otherwise not striking ; they ought to 

 '■ examined from time to time with very powerful microscopes, 

 for amongst these will be found, I believe, the new spectra 

 IW.h. 



There are other spectra, which, although they undoubtedly 

 belong to Class III.^', have not, it appears, reached their full de- 

 velopment. The least faint of these stars is that known as 

 7 Schj. Before my researches, nothing had been published 

 " regarding this spectrum except this short remark of D'Arrest, 

 "Irregular spectrum, probably type IV." ( Vierteljalirschrift der 

 Aslr. Ges. ix. Jahrg. p. 255). This spectrum presents the charac- 

 teristics of III./; very pronounced ; only band 5 is invisible, and 

 band 6 is so faint that at first sight the spectrum has not the 

 aspect characteristic of well-developed specti'a of this class. It 

 is for this reason that D'Arrest would say nothing positive re- 

 garding this star. If the spectrum of 541 Birm. represents the 

 first step in the passage of a star to Class lll.b, this star doubtless 

 represents the second step. Band 6 is the least developed of 

 the three principal ones. Although the spectrum of this star is 

 jjretty bright, band 5 is not visible, whilst band 4 is well visible, 

 md is also perceived in the spectrum of 541 Birm. 



In the spectrum of 19 Piscium (Fig. 4 on the map), which is 



me of the most magnificent in other respects, band 6 is still 



considerably fainter than the other two principal bands, whilst 



in that of 152 Schj. (Fig. 3) it is quite as pronounced as band 10, 



and almost as pronounced as band 9. This last spectrum is in 



;in advanced stage of development ; but in spite of that, band 4 is 



!>t stronger than in the spectrum of 7 Schj., and rather fainter 



•vxn that of 19 Piscium. The same relation is repeated in other 



^cctra of this class, so that sometimes band 4 is very visible in an 



therwiseless developed spectrum, but invisible in more strongly 



;uked spectra, and in the spectra of brighter stars of this class 



L'le are in the same way very faint band>, 7 and 8. But band 



i is in itself very pale ; it is the deep sodium line which makes it 



1 emarkable, and the bands 7 and 8 are probably only groups of 



Fraunhofer lines. 



It is therefore very probable that the more or less easy visi- 

 bility of these bands is no indication as to the phase of develop- 

 ment in which the star is. There is, on the contrary, reason to 

 believe that the strengthening of these lines, and also of the other 

 l)rincipal lines of the spectrum (except those of hydrogen, which 

 grow fainter during the passage of a star to Class III.) is a 

 I'rocess of relatively small importance which goes on whilst 



the star still undoubtedly belongs to Class W.a ; and even when 

 this is accomplished there is still nothing to show whether 

 the star will become Ill.rt or III./;, unless perhaps in those 

 which tend towards Class III. a, the line, or rather group of lines, 

 with wave-length 616. is very well marked, which seems not to 

 take place in the spectra Wl.l>. But in the stars which tend 

 towards the latter class the violet rays are already very much 

 absorbed, and the stars are therefore of a deep orange. 



If we pass on to consider the ulterior development of the star, 

 it is evident that as it cools farther it at length reaches a 

 temperature at which the carbon which must be present in 

 abundance, either in its atmosphere or under some form in its 

 photosphei^e.. can combine with hydrogen or some other element 

 to give the so-called hydrocarbon spectrum. After that, the 

 spectrum appears cut by a broad faint band with the wave-length 

 516/*, and by another still paler at 473/i, and the parts of 

 the spectrum beyond this are very faint. But gradually these 

 two bands increase in intensity, and at the same time the band 

 563,u is perceived, at first very faintly, and gradually becoming 

 stronger. At this stage the narrow band $^6|x is developed, 

 and finally the three principal bands are nearly of equal intensity, 

 and the spectrum shows all the characteristic details. It would 

 be useless to attempt to discuss the moment at which the 

 secondary bands in the red and orange make their appearance, 

 as no facts on the subject are known. 



It is doubtless very remarkable that in the spectra III.i^ no 

 trace of the carbon band with the wave-length 618 "J/U is seen, 

 which is so brilliant in Pliicker's tubes containing hydrocarbon. 

 This is, however, in perfect analogy with what is seen in the 

 spectra of comets, which owe their appearance to the same 

 carbon compound as the stellar spectra III./;, and there are 

 analogies also for the other bands. Thus the band 563,11 is 

 often very weak even in the bright comets, and the band in the 

 green is always the strongest both in comets and stars. The 

 band in the blue is sometimes pretty faint in cometary spectra, 

 whilst in the stars it is only a little fainter than the band in the 

 green ; but we must remember that it is situated in a very faint 

 part in the spectra of the stars. It is therefore very possible 

 that a little dimness should render the remaining light entirely 

 imperceptible. In this perhaps there is no diversity between 

 comets and these stars. The violet bands are very faint in 

 Pliicker's tubes, but strong in the flame of alcohol. A trace of 

 them has been seen in the spectra of the brightest comets. In 

 very brilliant, not too red stars Wl.l>, there is also a violet zone, 

 terminating at the wave-length 430^, of which there is a band 

 at the position of the first and the second of these bands in the 

 spectra of these stars. 



We will now pass on to consider the changes which take 

 place in stars of Class III. after their spectra have completely 

 developed. As the cooling goes on, they necessarily grow dim- 

 mer and dimmer, and at length become extinct. Either the 

 bands in their spectra must increase in width until at last the 

 shining intervals disappear, or else, the bands keeping their 

 same width, the whole spectrum grows fainter. Certainly we 

 see that there are stars whose bands are enormously broad, but 

 none the breadth of who-se bands surpasses that of the bright 

 zones. 



I think, therefore, we can hardly accept the first hypothesis, 

 but there are reasons which give very valuable support to the 

 second. We know that the weakness of the light in the solar 

 spots is, in the first place, caused by a general obscuration of 

 the spectrum, and that the enlargement of the Fraunhofer lines 

 has very little to do with it. Besides, I have exanv'ned, on 

 different occasions, between the maximum and the minimum, 

 the spectra of several variable stars of Class III., and found 

 that there was no widening of the bands sufficient to explain 

 the weakening of the stars. There is no doubt a remarkal le 

 analogy between the spectra of the sunspots and those of the 

 stars of Class III., and one which we have no cause to be sur- 

 prised at. For, on a'ccount of the relatively low temperature 

 of these stars, it is very probable that their surfaces are in great 

 part covered with formations similar to our sunspots, and the 

 absorption-bands found in their spectra are no argument against 

 this analogy. They prove only that chemical compounds may 

 be formed and maintained in the atmospheres of these stars, 

 which is not possible in !our sun, not even in the masses of 

 relatively low temperature of which the spots consist. 



Before laying down my pen I must remark that the induction 

 by which I arrived at these conclusions does not prove that the 

 spectrum of each star commences with Class I. and finishes with 



