February 4, 19 15] 



NATURE 



619 



proto-helium. A further inquiry on this subject is 

 necessary, and is under way. 



J6 38 +0 42. 4t4 46 



DLII 



ra 



na 



1 

 S 



Id 

 K 



IE 



I ■ II III 



n 



III! II 



n 



^7^ 



nni 



n 



n 



Fig. 2. — Showing how bright hydrogen lines of the stellar envelope make their appearance 

 superposed in the dark absorption lines produced by the subjacent atmosphere or photo- 

 sphere. This also occurs from the least refrangible side, extending to the blue end as 

 the intensity increases. 



It is clear from the above that bright Unes have 

 a way of their own, and that the absorption hnes 

 and the bright lines have to be studied separately. 



It must also be stated that in the majority of 

 non-variable stars with bright lines, the absorp- ; 

 tion lines have not yet been sufficiently studied j 

 to allow of their chemical classification. ] 



I next come to a different origin of bright lines j 

 studied in some variable stars. The variability | 

 of these is due to a brightening of the continuous j 

 spectrum and the appearance of bright radiations, \ 

 at maximum. I have previously (Bull. I.) referred j 

 to my hypothesis that these are produced by 

 collisions between meteor swarms. j 



The simplest case is represented by a swarm B ! 

 (Fig. 3) revolving round A and colliding at peri- 

 astron. The light curve of the variability thus 

 produced is represented in the middle part of 



\f Mtfu'atJijn 



The illustrations are "taken from Dr. Lockyer's 

 memoir on the variability of tj Aquila;. 



In a less simple case, there is a third 

 swarm, C, involved, with a much 

 COLi/MB/t shorter period than that of B. The com- 

 bined curve is no longer smooth. 



What, then, is the actual physical 

 condition of these two sets of stars with 

 bright lines? In order to answer this 

 question and to test the collision theory 

 in the case of the variables, we must 

 know whether their temperature is rising 

 or falling. 



A set of diagrams has been prepared 

 by Mr. N. K. Johnson showing the 

 Harvard and Kensington classifications, 

 the former, except at the bottom, corn- 

 et CENTAUni mon to the two arms of the temperature 

 curve, which is indicated by two vertical 

 lines and a connecting horizontal line 

 at the top.* 



To use these we may begin by taking 

 the case of the constant bright line 



RTAURI 



o PUPPI5 



v'PUPPfS 



i CENTAURI 



Fig. 3. — Orbits of two meteor swarms B and C round their primarj- A, colliding at periastron. 



Fig. 4. — Li'ght corves. B, produced by revolution of B : C, produced bj 

 revolution of C ; A, compoond carve produced by both revolutions. 



Stars, omitting the Wolf-Rayet stars. 

 Of the thirty-eight stars given in the 

 revised Harvard photometry, twelve, 

 indicated by crosses, are included in 

 Apastron my classification. All these are near 

 the top of the temperature curve. 

 There is a fair assumption then that 

 the other twenty-six stars may follow 

 suit — that is, that they probably 

 belong to the species which in- 

 clude the twelve; in the absence 

 of a complete classification, there- 

 fore, inquiries involving their physi- 

 cal state will be safer in accepting 

 this assumption than in neglecting 

 it. 



NORM.^N LOCKYER. 



Fig. 4, a rapid ascent followed by a slower 

 descent, and the curve is smooth. 



XO. 2362, VOL. 94] 



• The original temperature curve contained gaps below the Cygnian and 

 above the Sirian types for two other possible classes, but these are omitted 

 In the diagrams. 



