108 



KNOWLEDGE. 



[May 1, 1896. 



Antiiiphi/siail Journal. Mr. Roberts has chiefly investigated 

 8 Cephei as a good example of a short-period variable of 

 constant variation. 



Now we know, since J\I. Belopolsky's researches, that in 

 the case of S Cephei its variation is intimately connected 

 with a revolution in a very elliptical orbit. But as all 

 trustworthy observers of this star boar testimony of the 

 continuous waxing or waning of its light, every eclipse 

 theory of variation must be rejected. This theory would 

 also be inconsistent, as Mr. Roberts shows, with the 

 position of the maximum and minimum phases. Apart 

 from eclipses, there must be, therefore, another phenomenon 

 able to cause variability in close binary systems. 



That phenomenon, according to Mr. Roberts, is the 

 increase of heat at periastrou, the heat falling upon the 

 companion being then in the orbit of S Cephei nine times 

 greater than at apastron. Mr. Roberts supposes that thir, 

 greater heat near periastron causes the two stars to be 

 brightened up by an increase of temperature, the subsequent 

 cooling near apastron being the cause of the minimum. 

 But that explanation cannot be granted if the two stars 

 have some analogy with our sun. A sun cannot rise or 

 fall one or two magnitudes by changes of temperature in 

 the space of three days. 



But the brightening by periastric heat and darkening by 

 apastric cooling can be easily understood if we assume that 

 one of the two stars is a red star with an obscuring cloudy 

 atmospheric veil, alternately dissipated by the greater heat 

 near periastron and slowly recondensed again by the 

 subsequent cooling near apastron. Thus, near periastron 

 both stars would be visible, and near apastron (at least 

 clearly) only one. There, as well as in other red variables, 

 the change of brightness would be caused precisely as the 

 darkening of our sun is caused by even the smallest 

 clouds in our sky. If the hypothesis of Mr. Roberts is 

 thus amended it fits admirably well in my theory, as will 

 be seen in the following 



Classification of Variables. 



CI. A. — ^[ore or lef:s irregular Red ]'iiriahli's a7Hl Nova. — 

 Single stars internally always red, but covered by 

 cloudy atmospheric veils, alternately growing thicker by 

 cooling and then again dissipating by heat, which, due to 

 the intermittent formation of chemical compounds, is also 

 produced by cooling. Besides the somewhat capricious 

 changes of brightness thus caused (always attended with 

 a bright luminescence of the condensing atmosphere), 

 symptoms of a more regular periodicity can be expected if 

 phenomena analogous to those in the Classes B and C are 

 also working here. 



CI. B. — Pu'indar Yariahhx iritJi continuousvariation. — Close 

 binaries with very eccentric orbits. One of the stars is a 

 red star with a cloudy atmospheric veil (as in CI. A). This 

 veil is dissipated by the greater heat when the stars are 

 closest, and re-condenses by the subsequent cooling near 

 apastron. 



a. The plane of the orbit does not pass through the 



earth (S Cephei, X Sagittarii, T Vulpecuhc, etc., 

 etc.). 



b. The plane of the orbit passes through the earth. 



Apart from the periastric and apastric maxima and 



minima, secondary minima are produced by eclipses 



(i; Aquila?, R Sagittie, ^ Lyrre). 

 01. C. — Reyular Variables remainin<i im-ariably brijjht fur 

 most of the tints, with minima of short duration. — Close 

 binaries revolving fast in orbits whose planes pass through 

 the earth, and where eclipses of short duration are the 

 only cause of the variation. 



a. The binary consist of a bright and a dark star 



(Algol). 



/'. The binary consist of two equally bright stars 

 (Y Cygni). 



('. The binary consist of two differently bright stars 

 iZ Ilerculis). 



\'ariablos with long periods of invariable brightness and 

 darkness (as arc not uncommon in double stars), though 

 belonging, in a measure, on account of the cause of their 

 variability, to CI. C., will bo better classed in a separate 

 ('1. D. ; this class becoming thus finally intimately con- 

 nected with still another CI. Vj., containing the stars 

 which, as so many components in double stars, are 

 varying in colour. The description of the further particulars 

 of these Classes D and E may be postponed to a later 

 occasion. 



The case of r; Aquibr and R Sagitta^ has been studied by 

 Mr. iloberts (lor. rif., p. 289). In the case of ft liyrm (if 

 my explanation is right) the secondary eclipse minimum 

 must occur in the midst of the periastric maximum. The 

 cooler variable bright-line companion is then between us 

 and its primary. So we comprehend why (the major axis 

 of the orbit being nearly in our line of sight) the bright 

 lines of the companion and the corresponding dark lines 

 of the primary have always been observed as exactly coin- 

 cident during the secondary minimum, and why, before 

 and after that secondary minimum, these bright lines have 

 been seen displaced respectively to the red and blue 

 sides of their dark representatives. During the principal 

 minimum near apastron no eclipse minimum occurs, because 

 the eclipsed variable, then being in its darkest state, has as 

 little effect upon the brightness of the whole system as the 

 dark companion in the case of Algol. If, however, this 

 eclipse should be total, the bright lines of the companion, 

 which are never wholly extinguished, should temporarily 

 disappear, which phenomenon has been observed indeed 

 by Father Sidgreaves a few hours before the principal 

 minimum {Monthly Xot. 54, 2, p. 95). 



The red companion being relatively small, and com- 

 pleting, moreover, by its red, yellow, and green light, the 

 excess of blue and violet in its primary of CI. In, the 

 white colour of j5 Lyrse seems thus to be explicable. 

 The absorption-lines of the companion, though necessarily 

 hard to see on the relatively very bright and more con- 

 tinuous spectrum of the primary, may possibly be detected 

 afterwards. For, according to Prof. Vogel, " especially 

 good photographs have led to the supposition that with 

 sufficient dispersion the spectrum of /3 Lyrfe would be found 

 very rich in lines " (Antr. a7id Astrophijs., 1894, p. 302). 



The second investigation, about which I should like to 

 say a few words, is especially important in connection with 

 my idea suggested in Knowledge, December, 1895, that 

 all gaseous nebular spectia may be due to clusters wherein 

 the great majority of stars are Novas. That hypothesis 

 demands the existence of clusters wherein numerous stars 

 should be mutually equal, and very different from those 

 which are clustered in our neighbourhood. Though this 

 demand seemed to be in a measure supported by the fact 

 that stars of the same description show often a distinctly 

 gregarious tendency, and by Dunlop's and Sir .T. Herschel's 

 discoveries of clusters wherein all stars are respectively 

 blue or differently coloured, " like a superb piece of fancy 

 jewellery" {Ohs. at the Cape, p. 17, a. 102), it has been 

 quite recently much more strikingly verified by Prof. 

 Bailey's discovery, at Arequipa, of certain globular clusters 

 containing an extraordinary number of variable stars 

 (Astrophijs. Journ., November). In the cluster N. G. C. 

 6272 no fewer than eighty-seven stars have been proved to 

 be variable ; in the cluster 5904, at least forty-six. On 

 the other hand a similar examination of each of the 

 clusters 6218, 6397, GG26, 6705, and 6752 failed to detect 



