1855.] Sokbt on Detection of Blood- Stains by Spectrum-Analysis. 207 



my descriptions. Gas-lamp light is, if anything, better than daylight, 

 because the line C of Fraunhofer interferes with the correct deter- 

 mination of some facts. Turning a wire holding common salt into 

 the flame, the bright yellow sodium line is seen ; and one of the prin- 

 cipal lines of the micrometer having been made to bisect it, the wire 

 is turned out of the flame and the measurements made. 



Fig. 2 — Blood Spectra. 

 123456 78 



Since human life might depend on the accurate determination of 

 the facts, I shall describe all the characteristic peculiarities of the 

 spectra, so as to avoid, as far as possible, any serious mistake. If a 

 piece of linen | or xoth of an inch square, soaked with blood, and 

 quite recently dried in pure air, be digested in a few drops of water 

 in a watch-glass, it yields a solution, which, when introduced into one 

 of the cells just described, produces a spectrum like No. 2 in Fig. 2. 

 The blue end is quite absorbed, and so are two bands in the green, 

 but the whole of the red end is transmitted. When the solution is 

 stronger, the absorbed portion of the spectrum increases upwards, and 

 the dark bands in the green become broader, until the whole of the 

 light below D is absorbed, and merely a bright red remains above it. 

 If examined when the yellow sodium line is present a narrow dark 

 band is seen just above it, even when the spectrum is like No. 2, and 

 shows no such dark band in that position with the natural light. On 

 diluting the solution the bands in the green become more and more 

 narrow and faint, but do not disappear until it is so dilute that the 

 blue end of the spectrum is transmitted without sensible alteration. 

 Since the width of the bands varies with the strength of the solution, 

 the position of their edges is not constant, though that of their centre 

 is nearly so, and will be given in the following descriptions. How- 

 ever, it is somewhat difficult to determine it with great accuracy, on 

 account of the gradual shading off on each side, which of course can 

 be only imperfectly represented by a woodcut. Taking the whole 



