Ai'EiL 1, 1891.] 



KNOWLEDGE. 



73 



greatest indications of absorption in itS spectrum must be 

 actually emitting the greatest amount of light from its 

 photosphere. ■■ And, to say the least, it is arguable that 

 the fuller development of the metallic spectra in stars of 

 the Second Type shows a higher temperature, not a lower 

 one. It is conceivable, not to put it more strongly, that 

 an intenser heat may in these stars keep metals in the 

 state of absorbing gas, and at a higher atmospheric level, 

 which in Siriau stars are precipitated at or below the 

 photosphere, and so give no sign of their presence. At 

 all events, we can feel well assured that the temperature 

 of the Solar stars is very high, for not only do we find the 

 metallic lines in the visible spectrum, but we tind them 

 strongly marked and numerous far in the ultra-violet, a 

 proof that even in the reversing layer the temperature is 

 high enough to compel the metals whose gases compose it 

 to give out radiations of even the shortest wave-lengths, 

 and to give them out strongly and unmistakably. 



The other circumstance that has led to the belief that 

 the white stars are the hottest, the red the coolest, is that 

 whilst more than half the stellar giants are of the first 

 type, there are only two Ist-magnitude stars of the third 

 type, and the brightest of the fourth-type stars is not so 

 bright as the 5tli-magnitude. But, again, a closer con- 

 sideration of the facts leads us to a different conclusion. 

 Surely we have no right to assume that the stars are 

 equally distributed amongst the various Types. If they 

 really mark different stages in stellar evolution, some may 

 be much more quickly passed than others ; or it may be 

 that but few stars are yet old enough to have reached the 

 most advanced stage, or else that but few are so young as 

 not to have passed through at least one phase. There is, 

 however, a simple test which we can apply. If the orange 

 or ruddy stars are fainter on the average than the white 

 and yellow, then the further down in brightness we go 

 the larger will be the proportion of such stars observed. 

 But this does not appear to be the case. If we take those 

 stars of the Oxford Uranometria which are above the 5th 

 magnitude, and have been classified by Secchi, we find 

 them thus divided : — 



Type I. Typo II. Type III. 



Above 1st mag. ... ... 5 2 1 



Between 1st and 2nd ... 11 -1 1 



„ 2nd and 3rd .20 17 4 



Srdand-tth ... 50 54 2 



4th and 5th ... 85 29 2 



121 lOG 10 



The stn.rs of (hicm and of tlie Pleiades which have 

 spectra a little differing from the normal type, have never- 

 theless been included under Type I. Excluding these, and 

 such stars as were outside the general Oxford limit of 

 North Polar Distance (100°), and were only observed on 

 account of their special brightness, we find that for the 



* If the stars are all composed of similar materials (as has been, 

 perhaps, too hastily assumed), and their photospheres are clouds of 

 incandescent particles. The temperature of their photospheres 

 could not exceed the liighost temperature at which the most refrac- 

 tory materials would he driven into vapour, and all photospheres 

 would shine with equal brightness. That there are differences in 

 the apparent brightness, area for area, of stars can hardly bo doubted 

 from tlie facts wo have already learnt witli regard to the masses of 

 double stars ; such, for example, as Sirius and liis companion, and 

 the binaries of the Algol type. 



From tho facts we know with regard to our own sun we cannot 

 suppose that tho matter of gaseous stars is stratillcd at different 

 levels, as some theorists have too hastily assumed. For on the sun 

 tho continuous eruption of matter from below the photosphere would 

 sulliciontly mix the gaseous matter by transfer in mass, if we could 

 conceive of it as not mixed by diffusion. Any apparent stratilication 

 can only be duo to some vapours continuing incandescent at lower tem- 

 peratures than others. — A C. U.^nvaku. 



stars north of 10° S. Dec, and above the 5th magnitude, 



the numbers would run : — Type I. Type II. Type III. 



Above 1st mag. ... ... 2 2 1 



J5etween 1st and 2ud ... 8 4 



2Qd and 3rd ... IH 17 4 



,, 3rd and 4th ... 45 54 2 



4th and 5th ... 33 29 2 



lOG lOG y 



It would seem then that, within these Hmits, the first and 

 second types are about equally numerous, and each about 

 twelve times as abundant as the third. 

 , Comparing these numbers with the results of Vogel's 

 and Konkoly's spectroscopic surveys, which were carried 

 down to the 7i mag., we tind that they give out of 6,073 

 stars examined the numbers for each type as follows : — 



Tvpe I. Type II. Tvpe III. Tvpe IV. 



3,145 2,105 " 375 " 12 



besides 14 stars of the Orion variety of Type I. and 422, 

 that could not be properly classified under any of the above 

 heads. 



The comparison shows that, so far as these observations 

 go. Type I. has a small preponderance over Type II. for 

 the higher magnitudes, but that this preponderance ceases 

 between the 3rd and 4th magnitudes to become much 

 more marked as fainter stars are included. Three con- 

 clusions seem to be fairly deducible from this result. 

 First, that Sirian stars are more numerous on the whole 

 than Solar stars. Secondly, Solar stars are on the average 

 rather brighter than the Sirian stars, or else rather nearer 

 to us. Thirdly, that the average difference in the bright- 

 ness of the stars of the two Types is not by any means so 

 great as we should expect if either Type marked a stage 

 of very greatly superior temperature to the other. 



Of course the materials I have used are very incom- 

 plete and can only supply indications, and not proofs. 

 Still, it may be worth while to take the question a stao-e 

 further, and see what light paraUax determinations have 

 to throw on the question. 



Taking the table of parallaxes given in the appendix to 

 Miss Clarke's Systcin of the Stars — a book as valuable and 

 instructive as it is charming in style — we find 20 stars the 

 parallaxes of which are given, and which are found in 

 Secehi's lists of star Types. Adding Arcturus we have 

 21, of which nine are First Type, and 12 Second Type 

 stars. Employing the Oxford magnitudes and computing 

 the absolute light-giving power of each star, takmg Sirius 

 as our standard, we obtain the following table : — 

 Sirian Stars. Solar Stars. 



/3 Cassiopeia . 0-29 a Cassiopeiie . 1-41 

 a Persei . . 1-90 rj Cassiopeia' . t)-12 

 Sirius . . 1-00 y8 Andromeda' . 0-98 

 Procyon . . 0'56 Polaris . . 4-45 

 Kegulus . . 2-55 a Arietis . . 1-39 

 a Lyno . . 48-35 Aldebarau . . 1-68 

 a Draconis . 0-04 Capella . . 514 



a Aquilffi . . 0(51 Pollux . ;-j-!)2 



a Ce^jhei . . 1-GO >/ Herculis. . 0-15 

 TT Herculis . 0-19 



€ Cygni . . 0-50 

 Arcturus . . 147-00 

 It will be seen at once that two stars stand out from all 

 the rest ; \'ega amongst the Sirian stars benig noarlv six 

 times us bright as tho other eight taken together : and 

 Arcturus, amongst the Solar stars, more than seven times 

 as bright as all tho eleven others taken together. In 

 both cases tho parallax adopted is that of Dr. Elkin, 

 which for these two stars, and especially for \'oga, differs 



