1 873.] Comparative Vegetable Chromatologv. 453 



separating the alcoholic solution, and repeating the process 

 over and over again, with the addition of a little water each 

 time, a comparison of the spectra of the different portions 

 thus fractionally separated will often suffice to show whether 

 the original coloured solution was or was not a mixture, and 

 the extremes are sometimes different substances, in a more 

 or less pure state. Of course if the original had not been a 

 mixture, such a difference would not occur, unless some 

 decomposition took place, which could easily be detected. 

 There are, however, cases where different substances cannot 

 be separated in a satisfactory manner by such means, and 

 it would be almost impossible to study comparative vege- 

 table chromatology successfully, if light could not be made 

 use of as a reagent. On exposing to the direct rays of the 

 sun solutions of different colouring matters in bisulphide of 

 carbon or other solvent, some are rapidly decolourised, 

 usually, but not always, without the intermediate produc- 

 tion of any new coloured substance, whilst others fade very 

 slowly, some being changed by one kind of light, and some 

 by others. This decomposition usually depends upon the 

 presence of both air and light, and does not occur in the 

 dark when air is present, or in tubes quite free from air 

 when exposed to the sun. The result of this difference in 

 the behaviour of different substances is that, though they 

 cannot in some cases be separated by chemical means, one 

 may be entirely destroyed by exposure to the open sun, or to 

 particular rays which pass through coloured glasses, whilst 

 sufficient of the other remains unchanged to show its 

 characteristic properties in a satisfactory manner. By com- 

 bining the above-described method of fractional separation 

 with this kind of photo-chemical analysis, it is often easy to 

 unravel very complicated mixtures ; and I do not think that 

 anyone who had not tried this system of investigation 

 would be prepared to find how much may be effected 

 by such simple means. By adopting these several methods, 

 it is not only possible to detect a comparatively small 

 quantity of the more important constituents in complex 

 mixtures, but also to determine their relative amount in 

 different cases. This kind of comparative quantitative 

 analysis is of very great value in the present subject. It 

 does not consist in ascertaining the relative weight of the 

 different colouring matters in any one specimen, like the 

 ordinary sort of quantitative analysis, but in determining 

 the relative quantity of each colouring matter in two or 

 more different specimens.' In those cases where the con- 

 stituents can be more or less perfectly separated, the 



VOL. III. (N.S.) 3 N 



