contributions to the Spectral Analysis of the Stars. 367 



considerably weakened, that then the intensely luminous protu- 

 berances stand out from the illuminated parts of the corona 

 of the darkened sun. We may form an idea of the magnitude 

 of the necessary enfeeblement of the diffuse light of our atmo- 

 sphere, if we assume that the mean luminosity of the atmosphere 

 during a total solar eclipse is equal to that during an average full 

 moon. From my photometrical measurements* this luminosity 

 is 618,000 times less than that produced by the sun. Hence 

 the selective absorption of coloured media must stand in a similar 

 ratio to that of the homogeneous light of the protuberance, if, 

 as is attempted on various sides, we wished to make the protu- 

 berances visible without dispersion. 



On the other hand, the possibility of attaining this object by 

 the aid of the prism by dispersing the superposed atmospheric 

 light depends essentially upon the circumstance that this light 

 consists of rays of all refrangibilities, while that of the protube- 

 rances only consists of three homogeneous kinds of rays. 



I have in the following manner produced artificially the su- 

 perposition of a non-homogeneous mass of light over a body 

 shining with homogeneous light and bounded by sharp outlines. 



The wick of an alcohol-flame was impregnated with chloride 

 of sodium and chloride of lithium. At a distance of eighteen feet 

 from this flame, a piece of plate glass was so placed at an angle 

 of 45° to the direction of observation, that the reflected image of 

 a petroleum -flame at the side covered the feebly luminous alcohol- 

 flame, and by its considerably greater intensity rendered it quite 

 invisible. About a foot in front of the reflecting glass plate was 

 a small lens of 6 inches focus, which threw an image of the alco- 

 hol-flame upon the slit of the spectroscope. The latter was 

 fastened to the end of a spring about 10 inches long, by which, 

 removed from its position of equilibrium and left to itself, it 

 could for about five minutes be made to perform oscillations of 

 sufficient amplitude. 



The breadth of the slit was first of all so far diminished, that 

 when the slit was at rest the double line 1), and in comparison 

 feebly the lithium-line, appeared well defined in the field. 



When the slit was made to oscillate, these lines changed into 

 sharp images of the alcohol-flame, of which the two soda images 

 were about half covered. The apparent brightness of these three 

 images was considerably smaller than that of the bright lines, 

 and hence their prominence on the diffusely illuminated spec- 

 trum-ground smaller in the same ratio than that of the lines when 

 the slit was at rest. 



When now I applied the second of the above proposed me- 

 thods, and opened the stationary slit so far that the image of 

 * Photometrische Untersuchungen &c. p. 105. Leipzig, 1865. 



