102 



KNOWLEDGE 



[June 1, 1891. 



existence, but condition of development. For whilst 

 the total light-giving power of a star will depend 

 not merely on its " ago," but also on its size, so that 

 wo might expect both "old " and " young " stars to show 

 a great diversity of emissive power, it is not so with 

 its conditions of condensation. On the current theory 

 of gradual cooling and contraction the mass-brightness 

 would vary directly with the " ago," and we should expect, 

 if spectrum typo simply meant stage of development, that 

 there would be no very wide range of difference of mass- 

 l)rightness in any particular class, but that different types 

 would be clearly marked oft' from each other by very 

 different brilliancies per unit of mass. It should be as 

 easy to assign its spectrum to each star in the table, from a 

 consideration of the number deduced for it, as from actual 

 scrutiny by the prism. That the fact is so different shows 

 that we cannot accept such speculations as those of Vogel 

 and Lockyer as holding good generally ; spectrum type 

 may mean " age " in a number of instances, but it cannot 

 always do so. 



So far as the spectrum is an indication of the stage of 

 development attained it will undoubtedly be true, as I said 

 in April, that "theorists are probably right when they 

 have placed the Second Type stars as belonging to a later 

 stage than the First " ; but we must bear clearly in mind 

 that we have no right to assert that all stars, or even the 

 majority, pass thi'ough both stages. 



One most notable exception to the general greater 

 density of the Second Type stars must be mentioned. 

 This star, y Leonis, with a relative candle power per ton 

 of 92-99, stands out so distinguished for its superiority of 

 brilliancy over mass that I have not included it amongst 

 the other Second Type stars in the table. I think we are 

 warranted in treating this star, the co-efficient ot which is 

 half as large again as all the other 29 stars of its type put 

 together, as altogether exceptional. Nevertheless, its in- 

 clusion with the other members of the type would not have 

 materially altered the result. The average First Tj'pe 

 star would still have shown a higher mass-brightness than 

 the Second Type, whilst it would have become more, and 

 not less, obvious that condensation alone was not a suf- 

 ficient indication of spectrum. 



it is possible that the explanation of this peculiarity of 

 y Leonis may be that the Second Type is a stage ]iassed 

 through twice, once during the period of increasing tem- 

 perature, and later, after the First Type has been gone 

 through, during the time of decline. At present, however, 

 the star stands quite alone, and the only star which can 

 reasonably be suggested as also showing a similar extra- 

 ordinary relative brightness is Arcturus. If Arcturus 

 emits 147 times as much light as Sirius, it seems easier to 

 suppose that this corresponds to a greater emissive power 

 per ton, than to take the reverse view and assign it say 

 .lOO or (iOO times the mass of Sirius. 



It would bo an immense assistance to us in an enquiry of 

 this nature if we knew the details of the spectra of both 

 members of a large number of pairs. Unfortunately, there 

 arc special difticulties about the examination of the spectra 

 of dimblo stars, and. in consequence, it is a branch of study 

 which has been left practically untouched. We have, 

 however, many records of the rolouis of binaries, and, in 

 default of direct spectroscopic observation, they may be of 

 some ssrvice. Now these give us a most noteworthy 

 result. Where the two components of a binary are equal, 

 or nearly so, then both are nearly of the same colour like- 

 wise. Where one is much smaller than the other, then in 

 tlie great majority of cases the larger is yellowish, tlie 

 smaller blue. The blue star is never in such a case the 

 larger. Thus Mr. Lockyer gives, in bis " Moteoritic 



Hypothesis," tables of 120 binary stars, from Mr. Chambers' 

 revised "Bedford Catalogue." Of these 120 stars, the 

 colours of the companion are not given in 5 cases, leaving 

 115 to work with. Of the 115, -55 pairs show the same 

 colour — nearly always white to yellow^and the mean 

 dift'crencc in brightness is less than one magnitude, and 

 never exceeds two. In HO cases, the principal star is white 

 to yellow, and the small star blue, and for this class the 

 mean dilforcnce of brightness is two magnitudes ; and in 

 21 other cases the same arrangement of colours prevails, 

 but the mean difference of brightness is six magnitudes. 

 Five pairs only show the large star white or yellow, and 

 the small star red. 



The natural tendency would be to interpret the yellow 

 stars as being probably of the Solar Type, as many of 

 them certainly are, and the blue or purple stars as 

 strongly-marked specimens of the Sirian Type. Mr. 

 Lockyer does so, and so docs M. Eocques in a recent 

 paper on the "Ago and Colour of Stars." In one instance 

 at least we know that this interpretation is a correct one, 

 and though it is very likely that some of these small 

 deeply-coloured stars have spectra of a type as yet 

 unrecognised, yet the assumption that many are Siriau in 

 character is an extremely probable one. 



If so, four independent methods result in showing us 

 the Solar stars as on the average larger than the Sirian. 

 First, the Solar stars are represented in a higher propor- 

 tion amongst the third and fourth magnitudes than amongst 

 the sixth and seventh. Next, amongst stars of which wo 

 know the parallax, the Solar stars prove to have the highest 

 total light-giving power. Thirdly, the mean mass relative 

 to the light-giving power is much greater for Solar stars. 

 And now we seo tliat when the two components of a pair 

 are nearly equal tliey agree in colour, but when unequal, 

 then the smaller would appear to be of the Sirian, the 

 larger of the Solar Type. 



It seems to me that this curious subordination of the 

 blue stars in binary systems points to the principal star 

 having amassed to itself the bulk of the metallic con- 

 stituents of the nebula from which the system sprang ; or 

 else, if we adopt the idea of fission recently put forward by 

 Herr See, that the smaller body when thrown off consisted 

 mainly of the lighter elements, the heavier remaining in 

 the principal star. In other words, in these cases spectral 

 type depends upon original chemical constitution, and not 

 upon the stage of stellar development attained. 



A further circumstance which looks in the same direction 

 is the marked crowding of particular types in special regions. 

 Prof. Pickering has pointed out that whilst the Second and 

 Third Types are pretty evenly divided between the Galactic 

 and non-Galactic regions, two-thirds of the First Type stars 

 are foimd in the Milky Way, whilst the Fourth Type stars 

 notoriously afl'ect Birmingham's " Eed Region," in and 

 near Cygnus. Then again, the stars with spectra re- 

 sembling those discovered by Wolf and Fiayet, "//, though 

 they are 33 in number, lie within 10° of the Galactic 

 equator, the great majority within 2°. The Orion stars — 

 except Botclgeuse^again have a type of spectrum of 

 their own, scarcely found outside the limits of the con- 

 stellation. It seems only natural to infer that these 

 remarkable instances of unequal distribution of stars of 

 particular types are due not to the prevalence of one 

 particular evolutionary stage in certain districts, but to a 

 similarity in actual composition. 



Nevertheless, it will not do to entirely discard the idea 

 of successive stages, as evidenced by the difl'erent spectral 

 types. The table on previous page shows this. If type had 

 never anything to do with age, it is hard to see why 

 there should be such a marked superiority in mass-brigbt- 



