98 



KNOWLEDGE. 



[April 1, 1897. 



BOOKS RECEIVED. 



Wkifaier's Titled Persons, 1897. 2s. 6d. 



Diinamo Atfenilants and their Dipmmos. By Alfred H. Gibbings. 

 Third Edition. (Eentell.) Illustrated. Is. 



Celestial Motions. By W. T. Lynn, B.A. (Stanford.) Illustrated. 



28. 



Remarkable Comets. By W. T. Lynn. (Stanford.) 6d. 



Semariable Eclipses, By W. T. Ljnii. (Stanford.) 6d. 



Phifsical Tables. Prepared by Thomas Gray. (Smithsonian 

 Institution.) 



A Study of the Sky. Br Herbert A. Howe. (Macmillan.) Illus- 

 trated. 6s. 



HeUcs of Primeval Life. By Sir J. W. Dawson. (Hodder & 

 Stoughton.) Illustrated. 6s. 



Metals. By A. K. Huntington and W. G. Macmillan. (Longmans.) 

 Illustrated. 7s. 6d. 



The Journal of Malacolomi, If'SG. Edited by Wilfred Mark Webb, 

 F.L.S. (Dulau.) Illustrated. 



Proceedings of the Acadcmi/ of yatural Sciences. Edited by E, 

 J. Nolan. (Logan Square, Pbiladelpliia.) Illustrated. 



Chapters o;i the Aims and Practice of Teaching. Edited by Dr. 

 Frederick Spencer, M.A. (Cambridge L^niTersity Press.) 



Practical JSlectrical Mexsuremenis. By E. H. Crapper. (Whit- 

 taker ) Illustrated. 



Smithsonian Report, 1S04, of the U.S. National Museum. 

 (Government Printing OiBee : Washington). 



Our Secret Friends and Foes. By Prof. P. F. Franklaud, F.R.S. 

 (Society for Promoting Christian Knowledge.) Third Edition. 

 Illustrated. 3s. 



Molhs. Bv. W. P. Kirby. Allen's Naturalists' Library, Vol. IV. 

 Edited by R. Bowdler Sharpe. (W. H. Allen & Co.) lUustrated. 6s. 



Short Studies in Physical Science. By Yaughan Cornish, M.Sc. 

 (Samp.son Low, Marston, & Co.) Illustrated. 5s. 



Xew Thouqhts on Current Subjects. By the Key. J. A. Dpmc. 

 (Elliot Stock.) 



Librarq Year Sook, 1S97. Edited by Thouuis Greenwood. 

 (Cassell.)' Illustrated. 23. 6d. 



♦ • 



[The Editors do not hold themselyes responsible for the opinions or 

 statements of correspondents.] 



THE CHEMISTRY OF THE STARS. 

 To the Editors of Knowledge. 



Sirs, — In a note on my article on "The Chemistry of 

 the Stars" in the last number of Knowledge, Mr. Maunder 

 takes objection to the conclusion that spectral differences 

 are due to different stellar temperatures, seemingly on the 

 ground that it necessitates the "inadmissible" assumption 

 that the chemical elements are distributed in like pro- 

 portions throughout all the heavenly bodies. So far from 

 regarding the similar chemical composition of all stars as 

 an assumption, I look upon it as a result, which, if, perhaps, 

 not yet absolutely demonstrated by the available evidence, 

 is at least not in contradiction with anything that we 

 know concerning stellar spectra. 



In the first place, stars with spectra identical with that 

 of the sun are found in all parts of space — for they are 

 seen in every direction and are of all magnitudes — and in 

 all such cases the evidence is complete, not only that 

 similar elements are in question, but that the absorbing 

 regions of these stars have precisely the same percentage 

 chemical composition. 



Again, if the variety of stellar spectra is due to funda- 

 mental differences of chemical composition, one would 

 expect to find spectral differences among stars which are 

 at the same temperature. Now, the relative temperatures 

 of the stars (that is, of their most effective absorbing 

 vapours) can be determined with considerable certainty : 

 by the study of the magnesium lines, as Scheiner has 

 shown ; or, better still, by the varying representatives of 

 iron at different temperatures, as Lockyer has more 

 recently demonstrated. When this test is applied, the 

 verdict is that stars with equal temperatures do not exhibit 

 any such differences among themselves as to suggest 

 varying chemical compositions. 



In these inquiries allowance must be made for the fact 

 that spectroscopic invisibility does not necessarily imply 

 non-existence (as witness the non-reversal of the helium 

 lines in the solar spectrum), and for the changes in metallic 

 spectra at different temperatures. The importance of the 

 latter point is strongly emphasized by the recent work of 

 Lockyer, showing, with great probability, that the two 

 thousand lines of iron seen in the arc spectrum are reduced 

 to nine at the temperature of the hotter stars.* Bearing 

 these points in mind, as well as the fact that a particular 

 stellar spectrum is always associated with a particular 

 temperature, the action of different temperatures upon the 

 same mixture of chemical substances appears quite com- 

 petent to produce the observed differences in the spectra 

 of the stars. 



In only one sense, it seems to me, can it be held that 



the percentage compositions of stars are different, and that 



is to admit that in the hotter stars the chemical elements 



are dissociated, some at one temperature and some at 



another, so that it would be more correct to distinguish 



between elements and their germs. Percentage differences 



of this kind, however, are not fundamental. , _ 



A. Fowler. 



[In replying to Mr. Fowler, in order to render perfectly 

 distinct the point at issue between us, I would like to 

 quote the conclusion of an article appearing in Knowledge 

 for June, 1891. " The types of stellar spectra are not to 

 be explained along one line only, but along two : they do 

 not always denote the phases attained by the stars in their 

 development ; they often indicate, instead, radical differ- 

 ences of constitution." I would also repeat what I stated 

 in commenting on his article in the March Number, that 

 his view of the meaning of the relation between the different 

 kinds of stellar spectra " is certainly one interpretation 

 which can be put upon that relation, but not the sole one." 

 I do not dispute for a moment — it would be impossible to 

 do so — that difference of temperature condition is in very 

 many instances the true explanation of difference of stellar 

 spectrum. I only challenge it as the sole explanation and 

 as being universally applicable. I cheerfully concede that, 

 in some instances to which Mr. Fowler alludes, the demon- 

 stration is as complete as we can reasonably expect with 

 our present knowledge. 



I wish to emphasize this point, as it seems to me that 

 Mr. Fowler has fallen into the mistake of supposing that 

 if he can prove spectral type to be an indication of " star 

 age " in several cases, it must necessarily be so in all. 

 The reverse is rather the case. If the hypothesis of the 

 exact equal distribution of the elements is upset in a single 

 instance, it is rendered untenable as a postulate altogether. 



I think there are facts which must lead us to this con- 

 clusion. I can but briefly summarize some of them here. 



If temperature change be the sole cause of difference ol 

 type, then it follows that every star must in its turn pass 

 through every single variety of type. None whatsoever 

 can be omitted, no matter how unique and bizarre — unless, 

 perhaps, we except the type of liighest temperature, uni- 

 versally supposed to be that of the Sirian stars. To take 

 a very extreme case : Prof. Pickering has just reported to 

 us the discovery of a star of a perfectly new type, Xeta 

 Piippis. On Mr. Fowler's assumption, we must suppose 

 that every star in the sky must have shown this curious 

 triple character in the past, or must show it in the future. 

 And the new rhythmical series must belong to some sub- 

 stance present in the same proportions in every single 

 celestial body, but the spectrum of which is developed only 

 at the precise temperature which Zeta Puppis and no other 

 known star has reached. 



» Nature, March 11th, 1897. 



