November 12, 1914] 



NATURE 



28 



Cannon, who employed H.i., but some groups 

 were discarded and stars between F and G, for 

 instance, were classed as F5G or F8G. We may 

 mark this as H.3. 



The work of Harvard is monumental ; tens of 

 thousands of stars have been classified so far as 

 small dispersion can do it, but obviously a 

 straight line, instead of a curve of ascending and 

 descending temperatures, has been followed in the 

 three classifications used one after the other, and 

 in all, if my views are sound, stars of vastly 

 different constitution, sparse swarms and con- 

 densed stars, are called by the same name. 



A subject of great regret is that, as this is the 

 only large scale classification extant, it is now 

 being used to deal with questions in which it is 

 vital that stars differing in physical constitution 

 shall not be classed together; in which such a 

 mixing invalidates the conclusions arrived at in 

 such inquiries. 



In the last authoritative book published on 

 these matters Prof. Eddington writes : — " From 

 time to time there are indications that the Draper 

 classification has not succeeded in separating the 

 stars into really homogeneous groups. Accord- 

 ing to Sir Norman Lockyer there are stars of 

 ascending temperature and of descending tem- 

 perature in practically every group ; so that, for 

 example, the stars enumerated under K are a 

 mixture of two classes, one in a very early, the 

 other in a late stage of evolution." 



Prof. Russell, a diligent worker on the charac- 

 teristics and spectra of stars, has written as 

 follows :■ — 



" I have endeavoured ... to set before you the 

 present state of knowledge concerning the real 

 brightness, masses, densities, temperatures and 

 surface brightness of the stars, and to sketch the 

 theory of stellar evolution to which the study of 

 these things has led me. This theory is inconsis- 

 tent with the generally accepted view. Its funda- 

 mental principle is identical with that of Lockyer's 

 classification." 



Prof. Russell objects to the principles on which 

 1 have placed stars in the different groups but that 

 is another, and less important, story, which will 

 be dealt with later. 



Prof. Ludendorff ^ has recently provided a case 

 in point; he has discussed Prof. Campbell's results 

 on radial velocities and stellar types based upon 

 he Harvard classification. Difhculties were met 

 with. Dr. Ludendorff points out that these are 

 avoided and that a distinct differentiation of the 

 ^■elocities is obtained if the stars are arranged 

 in the order of ascending and descending tem- 

 peratures. Of the sixty-three stars selected by 

 Campbell, eighteen are contained in my classifica- 

 tion; of these ten are on the ascending arm, and 

 the velocities are all 4- ; the remaining eight are 

 on the descending arm, and the velocities, with 

 one exception, are — . The exception is a star 

 \ ery near the top of the curve. 



In arranging the programme of work for the 

 Hill Observatory, the first place has been assigned 



5 AstroHomische Nachrichten, No. 4847, vol. cxc, p. 193. 



XO. 2350, VOL. 94] 



to the continuation of the catalogue of the bright 

 stars classified on my hypothesis (of two groups 

 of stars, one increasing, the other decreasing 

 their temperature), so that the 470 we already have 

 may be as .soon as possible increased to icoo, 

 which will give us a broader base. 



I have also (X'cupiecl some time in considering 

 stars varying from the normal condition dealt 

 with in my published classification. These are : — 



Stars with bright lines. 



Variable stars with bright lines. 



\'ariable stars without bright lines. 



Double stars without bright lines. 



The object of my inquiry was to ascertain 

 whether further generalisations were suggested 

 as regards the arm of the curve on which these 

 stars should be most probably found, irrespective 

 of their chemical constitution on which my classi- 

 fication was founded. 



This further inquiry was based on my hyjx)- 

 thesis^ that the bright lines in variable stars were 

 due to collisions between meteor swarms. Duner 

 has shown that in Class 1 1 la (Antarian), of 297 

 stars 44 were variables, that is, i in 7. I found " 

 in 1894 that in j8 Lyrae two bodies resembling 

 Rigel and Bellatrix were involved. This lands us 

 in the Crucian stage, nearly at the top of the 

 ascending arm. 



When we deal with variable stars without bright 

 lines (eclipsing variables), the light curs'es show 

 that we are dealing no longer with swarms, but 

 with bodies with discs and therefore with photo- 

 spheres, and therefore again approaching the solar 

 condition. Now the best known variable of this 

 class is Algol, the type star of one of my groups 

 nearly at the top of the descending arm, nearly 

 on a level with the Crucian group on the other. 



From, these facts we learn that bright line 

 variables with the Harvard Classification, B. .A. F- 

 K. M. should be stars increasing their tempera- 

 ture, while eclipsing variables marked B. A. F. 

 K. N. should be decreasing their temperature. 



The result of the preliminary inquiries so far 

 has been very encouraging. 



Stars with bright lines have been found in 

 species represented on the ascending arm of 

 the curve, and none on the descending side. 



All variable stars with bright lines so far studied 

 belong to the ascending side, eclipsing stars which 

 do not show bright lines to the descending arm. 



A great majority of the double stars without 

 bright lines so far considered belong to the de- 

 scending arm. 



Details on all these points will be given when 

 the research is further advanced. 



It may be remarked that the danger of using 

 the Harvard classifications is not the same for 

 all the groups. 



The normal stars which differ most in their 

 physical state are the Antarian (ascending) and the 

 Piscian (descending). These are differentiated by 

 the adjacent letters M and X (A and Z would 

 have been better). 



One of the finest memoirs which has ever been 



6 Proc. Roy. Soc, iS88, p. So. 



Proc. Roy. Soc., vol. xH.. p. 27S. 



