I.] PRINCIPLES AND METHODS. 5 



the bluish- white stars, the white stars, the yellow, red and blood-red 

 stars to indicate a decreasing order of temperature.* 



We shall not go far wrong in supposing that the star with the most 

 intense continuous radiation in the ultra-violet is the hottest, inde- 

 pendently of absorbing conditions, which, in the absence of evidence to 

 the contrary, we must assume to follow the same law in all. 



An inquiry into the facts placed at our disposal by stellar photo- 

 graphs, shows that there is a considerable variation in the distance to 

 which the radiation extends in the ultra-violet, and that the stars can 

 be arranged in order of temperature on this basis. 



Judged by this criterion alone, some of the hottest stars so tar ob- 

 served are 7 Orionis, fOrionis, aVirginis, y Pegasi, ?? Ursse Majoris, 

 and A. Tauri. Of stars of lower, but riot much lower, temperature than 

 the above, may be named Kigel, f Tauri, aAndromedae, /3Persei, 

 a Pegasi, and j3 Tauri. 



In this way spectrum analysis helps us with regard to temperatures, 

 both on the earth and in the heavens. 



Discontinuous Spectra with Bright Lines. 



Let us next pass from a solid which retains its incandescence like 

 platinum wire without melting, or a liquid which retains its incan- 

 descence, like molten iron, without volatilising and see what happens. 

 We have found that when the light entering the slit consists of every 

 colour and every tone, we have a continuous band of colour. If there is 

 any defect in the light we must have a discontinuous one, for the reason 

 that an image of the slit cannot be produced in any particular part of 

 the spectrum if there be no light of that particular colour to produce it. 



There are many artificial flames which are coloured, and if their 

 light is analysed in the same way as the light of the candle, a perfectly 

 new set of phenomena present themselves. 



Let us again make use of our improvised spectroscope, and allow 

 the needle to be illuminated by the flame of a spirit lamp into which 

 salt is gradually allowed to fall we see at once why the flame is 

 yellow. It contains no red, green, blue, or violet rays, so that we 

 should not represent the spectrum by 



* On this point 1 wrote as follows in 1892 : " An erroneous idea with regard to 

 the indications of the temperature of the stars has been held by those who have 

 not considered the matter specially. It has been imagined that the presence of the 

 series of hydrogen lines in the ultra-violet was of itself sufficient evidence of a very 

 high temperature. The experiments of Cornu, however, have shown that the 

 complete series of lines can be seen with an ordinary spark without jar. Hence 

 the high temperature of such a star as Sirius is not indicated by the fact that its 

 spectrum shows the whole series of hydrogen lines, but by the fact that there is 

 bright continuous radiation far in the ultra-violet. 



