Method of Registering Absorption-Spectra. 315 



an error, as it must be remembered that the shades produced 

 photographically have not the same gradations as the inten- 

 sity of light, as Bunsen and Eoscoe first showed. 



Fig. II. shows the absorption of the violet glass, on the pre- 

 sumption that the intensity of the radiations is equal through- 

 out the spectrum, an assumption which is very generally 

 made. 



I have found that it is convenient in taking these spectra 

 to modify this method. The absorption produced by potas- 

 sium chromate takes somewhat of a wedge-form, shading off 

 from darkness in the violet to total transmission at the least- 

 refrangible end of the spectrum. If a dilute solution of this 

 substance be interposed in each case between the source of 

 light and the slit for half the time of exposure, we have an 

 impression of the spectrum the varying intensity of which is 

 less marked than if such an artifice be not employed. 



I may here remark incidentally that the passage of light 

 through an aqueous solution seems to interfere very little 

 with the intensity of the photograph at the least-refrangible 

 end. I had looked for a marked diminution, but have scarcely 

 noticed it. 



In photographing these absorption-spectra the source of 

 light should be brilliant : sunlight, the image of the incan- 

 descent points of the electric light, or the oxy hydrogen light, 

 may all be used ; but I prefer sunlight, as we are enabled by 

 the Fraunhofer lines to fix the locale of the absorption-bands 

 more readily than with the other two. 



Another application of this method is to the solar spec- 

 trum itself. Researches have shown that the bright-line 

 spectra of incandescent compound bodies should lie in the 

 least-refrangible end of the spectrum, and that to discover 

 these a search must be made in these regions. As far as the 

 visible spectrum is concerned such a search has been made ; 

 but we have yet to examine those regions which are invisible. 

 At a low temperature it is quite possible that the compound 

 bodies should give off vapours of the compound, whilst at high 

 temperatures, such as that of the electric arc, they are pro- 

 bably dissociated. If, then, we wish to ascertain the exist- 

 ence of such compounds in the photosphere, we are driven to 

 compare the solar spectrum with the bright-line spectra of the 

 various compounds when heated at such low temperatures as 

 those of the ordinary colourless gas- or spirit-flame. To pho- 

 tograph portions of such spectra (even the most "actinic" 

 region of the spectrum) is a feat of uncommon difficulty; and 

 it would require hours, I might say days, of exposure to im- 

 press lines in the red-region. Such an attempt would be 



