484 



NATURE 



[August 21, 19 19 



for the 



l^OTES ON STELLAR CLASSIFICATION. 



IN Nature, December 23, 1915, and in the third 

 Bulletin of the Hill Observatory, I referred to 

 the shape of the temperature curve which I had pub- 

 lished in connection with the meteoritic hypothesis, 

 and I pointed out that if we could deal with a large 

 number of stars, a generalised temperature curve 

 might be placed before us by considenng the 

 number of stars in the various groups '— ^^- 

 reason that the longer a star re- 

 mained at about the same tempera- 

 ture, the larger would be the number 

 of stars in that group, while a rapid 

 rise of temperature would reduce the 

 number. I gave the curves thus pro- 

 duced by discussion of the stars in- 

 cluded in the catalogue of the 470 

 brighter stars published in 1902, and 

 in the later catalogue of the 354 less 

 bright stars catalogued at the Hill 

 Observatory. 



In order to carry the inquiry one 

 step further, I now reproduce these 

 two curves, together with a third 

 (Curve 3) based on the catalogue of 

 287 stars, the result of still more 

 recent work at the Hill Observatory. 



One of my chief objects in plotting 

 this third curve was to see whether 

 its shape agreed with the two former 

 ones, because the more the curves 

 based on different catalogues agree, 

 the more they may be accepted as a 

 basis for consideration. 



It will be seen that the third curve 

 follows suit with the first and second. 

 Kinks occur in practically the same 

 positions both on the ascending 

 and descending arms of the curve. 

 The main difference is that the apex 

 of the curve occurs later in the case 

 of the hotter stars than it does in 

 either of the others ; but the remark- 

 able verticality of the curve near the 

 middle of the ascending side is 

 common to all, and, indeed, is one of 

 the most striking features. 



If the similarity of the three curves 

 obtained from different data may be 

 taken as suggesting a probability 

 that the classification on which they 

 are based does really provide us with 

 homogeneous groups of stars on both 

 sides of the curve, several interest- p 



ing inquiries are suggested. 



Supposing that the stellar systems with which 

 we are dealing were of very recent origin, it is 

 clear, if the meteoritic hypothesis is true, that 

 the stars will all be found ^n the ascending arm. 

 If, on the contrary, the systems are very old, and 

 there are no recent formations, it is the descend- 

 ing arm into which they will be crowded. 



If my classification embracing high and low 

 temperatures really does provide us with homb- 

 geneous groups of stars, some hotting, some 

 NO. 2599, VOL. 103] 



cooling, and if such a result proceeds from either 

 a simultaneous or a continually acting formation 

 of star groups, a break in the series can only be 

 due to the cause I have already considered in 

 Bulletin IV., a more rapid change of temperature 

 giving an accelerated stellar change at one point 

 of the curve. 



But on the supposition that neither a simul- 

 taneous nor a continually acting formation took 



Temperature Curve , I , 



1 TM6 SPtCTRA or THE 471 STARS CLASSlFlfff 

 SOLAR Physics OBSERVATORY, south KtNSiNCToN. 



PERCENTAGE Of STARS IN RESPECTIVE SPECTRAL CROUPS 



Temperature Curve, 2 , 



Based on nt Spectra oFTMt 3S4 stars classified «t the 

 Hill OBSERVATORY. sioMooTH, 6 c»rAU)cuEo IN bulletin n'IIi. 



tLNlTAMIAN 



Percentage of stars in respective SpectralCroups 

 Temperature Curve 



Polar IAN . 

 AlOEBARIAN. 



PERCENTAGE OF ST ARS 



RESPECTIVE SPECTRALCROUPS. 

 numbers of stars and temperatures. 



place, then we should expect breaks or a break in 

 the curve. Supposing one break, we should be 

 dealing with two groups of stars representing the 

 old and the newer formations, or let us say old 

 and new star systems. If this be conceded, the 

 classification lands us in a new region of thought 

 which it is important to study, and the vertical 

 part of the curve may be taken as indicating the 

 locus of the cessation of the old system and the 

 advanced guard of the new. 



