RECENT ADVANCES IN SCIENCE 439 



i.e. they passed from blue to red as they condensed. The 

 alternative theory, developed by Hertzsprung and Russell, 

 which is gradually gaining acceptance, supposes that in the 

 early stages of a star's evolution, it is very diffuse, and has a 

 spectrum of the red type ; that, as the star condenses, it 

 becomes brighter and bluer up to a certain limit, after which 

 it commences to decrease in brightness, and again to become 

 redder. The turning-point depends upon the star's mass. 

 The existence of red stars of very great absolute luminosity, 

 and the non-existence of blue stars of very small absolute 

 luminosity was an important point in favour of this theory. 

 The result of Shapley seems conclusive in its favour, for if 

 the red stars could come only at a late stage in the evolution, 

 their densities must necessarily be very considerable. Further 

 systems are to be examined to strengthen the evidence. 



Determination of Large Proper-motions. — The stereo-com- 

 parator with its blink microscope has been applied by R. T. A. 

 Innes (Circular No. 25 of the Union Observatory of S. Africa) 

 for the purpose of discovering proper-motions. For many 

 purposes in astronomy it is desirable to know all the large 

 proper-motions in a certain region of the sky. The advantage 

 of the blink method is that it automatically separates the 

 stars with large motion from those with small motion. With 

 other methods this is not possible, and much unnecessary 

 labour is involved in seeking for the large motions. The case 

 is clearly put by Innes himself : " This blink method of deter- 

 mining proper-motions appears to be superior to all others in 

 its simplicity, certainty, and inclusiveness. The meridian- 

 circle method, with many costly observations and many years' 

 interval, will yield a few proper-motions of the brighter stars ; 

 the great ' Carte-du-Ciel ' Catalogue, when it is duplicated 

 in the far distant future, will by prodigious calculations 

 extend the list to perhaps the i2th-magnitude stars. But the 

 blink method only requires plates separated by moderate 

 intervals of time, and its work at once goes to the very limit 

 of photographic magnitude. It ignores what neither the 

 meridian instrument nor the ' Carte-du-Ciel ' ignores, namely, 

 the stars of no proper-motion ; it rejects the chaff automatically. 

 It will give in a few years a knowledge of the proper-motions 

 of faint stars such as without its help seemed to be for ever 

 beyond our reach." The method consists in viewing together 



