1 68 



NATURE 



[August 4, 1923 



28 000. 



_* X. Persel 



25000" 



22.000L 



20.000 



.*YPe9asi 



The temperatures of the photospheres arc deduced 

 from the distribution of 

 energy in the continuous 

 spectra. Laboratory ex- 

 periments on black-body 

 radiation show that the 

 spectrum of a perfect 

 radiator at uniform tem- 

 perature is characteristic of 

 the temperature. Wien's 

 displacement law states 

 that the wave-length of 

 the radiation carrying the 

 maximum amount of 

 energy is inversely pro- 

 portional to the absolute 

 temperature of the source 

 of radiation, and Planck's 

 radiation formula ex- 

 presses, with great accur- 

 acy, the distribution of 

 energy throughout the 

 spectrum. Assuming that 

 the photosphere of a star 

 bears the same relation to 

 its continuous spectrum as 

 does the equivalent of a 

 black body in the labora- 

 tory, the photospheric 

 temperatures are found. 

 Wilsing and Scheiner, of 



I8 500L 

 17800. 



ISOOO^ 



13800- 

 13300- 



12000- 



10400 

 10000^ 



5 520- 



50009 



4 700 



•k S Persei 

 -•^ Ariefis 



8200. -• Polaris 

 5 000-ffi^QL Persei 



1980- 

 1750- 

 1510- 



nso- 



-♦ at Andromedae 



-*T Cassiopeiae ^ , , >t , 



.-* Alqol ( S Persei ) Potsdam, and Nordmann, 

 of Paris, have used the 

 method, and Sampson has 

 .* Vega ( a Lyrae) recently shown that, by 

 employing a photo-electric 

 cell as an energy detector, 

 the accuracy of the mea- 

 surements may be greatly 

 increased. 



The measurements give 

 no indication of the region 

 of the star which has the 

 calculated temperature. 

 The results are spoken of as 

 the " effective " tempera- 

 tures, and are generally 

 assumed to characterise 

 the stellar surfaces, im- 

 mediately be- 

 neath the atmo- 

 spheres. Two 



L*p Persei " remarks might 



^yilil_*^Betelgeuse(x Ononis) be made con- 

 cerning them. 

 First, — granting 

 for a moment 

 the validity of 

 the method of 

 measurement, 

 they represent 

 minimum tem- 

 peratures only, 

 for if the stars 

 are not perfect radiators, their temperatures_must be 



NO. 2805, VOL. 112] 



_Sun 



a Aurigae 



A' 6 Andromedae 

 X^-k Aldebaran (a Taun) 

 Crater i&) of the electric ore 



3 700, 

 3 600- 

 3 500 

 2870 



2700-Jlt — ii w ucLci 



2680" '^flcetulene Flame „ ., 

 2170-111— AVCrW/? name * RY Draconis 

 tielting point of platinum 



copper 

 I Sifyer 



fead 



600- 

 273- 



0% 



Fig. 1. — Scale showing, in absolute centigrade 

 degrees, the temperatures attained in certain 

 terrestrial processes compared with the effective 

 temperatures of representative stars. 



higher than the calculated ones. Second, — it is a 

 somewhat dangerous assumption that the resultant 

 radiation from a globe of gas, perhaps millions 

 of miles in depth and varying in almost every 

 physical quality from point to point, will give a 

 spectrum comparable with that of a thin solid surface 

 at a uniform and probably \tvj much lower tempera- 

 ture. We know practically nothing as yet of the 

 processes of production of continuous spectra. We 

 have no means of distinguishing one such spectrum 

 from another except by measuring the distribution 

 of energy in it ; yet it is certain that there may be 

 profound differences in the modes of origin. The 

 continuous spectra of a cold fluorescent body, of an 

 electric glow-lamp, of hydrogen radiating also the 

 Balmer series — here at least are three spectra which 

 probably have nothing in common except their appear- 

 ance. The stellar nuclei of planetary nebulae, again, 

 give spectra which suggest the operation of the classical 

 laws of radiation rather than those of the quantum 

 theory, unless the stars have temperatures so high that 

 no one is prepared to accept them. 



It is noteworthy, however, that the atmospheric and 

 photospheric temperatures, estimated by totally 

 different, and at best approximate, methods, are of the 

 same order of magnitude. Fig. i ^ shows, on a thermo- 

 metric scale, the range of temperatures covered by 

 present measurements. Temperatures have been 

 measured at almost all points intermediate between 

 the absolute zero and the temperature of { Persei. The 

 cores of the stars, according to Eddington's theoretical 

 researches, reach temperatures far too high to appear 

 on the scale. It is probable that there are bodies in 

 the universe at all temperatures between absolute 

 zero and 20 million degrees centigrade or higher. 



Whatever may be said of the absolute accuracy 

 of stellar temperature measurements, it is scarcely 

 questionable that they show the true order in which 

 the temperatures are arranged. There is no doubt 

 whatever that Vega is hotter than Aldebaran in corre- 

 sponding regions. Consequently, if the order of stellar 

 evolution can be established from other data, it becomes 

 possible to determine the changes of temperature of a 

 star throughout its life. Russell's well-known theory 

 of evolution takes the order of increasing density of a 

 star to be its order of development : contraction is a 

 continuous process from childhood to old age. This 

 implies that a star passes twice through the same 

 series of spectral types, and therefore through the same 

 series of temperatures. Beginning as a huge, rarefied, 

 cool mass of gas, it contracts and becomes hotter until 

 a stage is reached when it is too dense to obey the laws 

 of a perfect gas. The temperature then soon reaches 

 a maximum and begins to fall — contraction, however, 

 continuing, though at a slower pace — and the star 

 retraces its path through the sequence of spectral 

 types which it traversed on its upward journey. While 

 the temperature is rising, the star is a " giant," and after 

 it begins to fall the star becomes a " dwarf." The 

 career of a typical star, with time as abscissa and 

 temperature as ordinate, is pictured in Fig. 2 : con- 

 tinuous contraction is indicated by the decreasing 

 diameter of the circles representing the star. 



' The diagrams illustrating this article are adapted, by kind permission 

 of Dr. Charles Nordmann, from an article by him on " La vie et la mort des 

 etoiles," which appeared in V Illustration of April 7, 1923. 



