FOOD-COLOEING SUBSTANCES. 17 



Sec. 17. — The Sudans and similar dyes of this group may be further separated with 

 gasoline and 90 per cent methyl alcohol. They may be separated from troublesome 

 accompanying oily impurities, by shaking out the gasoline solution with 85 per cent 

 phosphoric acid to which has been added from 10 to 20 per cent of concentrated sul- 

 phmic acid, though some dye is likely to be destroyed in this treatment. 



Like the colora described in sections 15 and 16, these dyes are almost quantitatively 

 removed fi-om gasoline solution by 90 per cent phenol. The phenol may, of course^ 

 be dissolved in alkali and the color again taken up with ether or gasoline, effecting a 

 eepai'ation from some impurities. 



Sec is. — When Naphthol green B is present the salt solution should not be made 

 strongly acid since small amounts of tliis dye decompose quickly in acid solution. 

 When its presence is suspected, the neutral salt mixtm-e may be first extracted with 

 dichlorhydrin washed once with benzene to remove the dissolved solvent, made half 

 normal with hydrochloric acid, and then shaken out with anilin. (It is best to add 

 the solvent before the acid.) From the anilin solution the dyes may be fractionated 

 by shaking out with 5 or 6 per cent salt solution which contains hydrochloric acid 

 varying from fourth-normal to sixty-fourth-normal. 



No outline in the form of a key can be so useful as a table of solu- 

 bilities. The solvents and the order in which they are to be used 

 obviously may be varied when the analyst has information regarding 

 the source and appearance of a sample. For example, in the case of 

 the red color solution obtained from commercial cocktail cherries 

 known to be ordinarily colored with Erythrosin, it would be better 

 to make such solution acid and shake it out with ether first, the 

 complete extraction of the dye indicating at once the absence of all 

 excepting one group of colors. 



In the examination of dye solutions the analyst of some experience 

 often will prefer (especially with colors extracted from acid solu- 

 tions by amyl alcohol) to wash the organic solvent extract with 

 successive small portions of water instead of with hydrochloric acid 

 or other aqueous solvent of definite concentration. In the case of 

 an acid amyl alcohol extract much hydrochloric acid is taken up by 

 the solvent along with the coloring matter. The acid is washed out 

 rather slowly, and as a result of this gradually decreasing acidity of 

 the washing coloring matters of different solubility will appear in 

 maximum amount in dilTcrent fractions, the order being apparent 

 from the solubility tables. With binary mixtures usually some of 

 each c(jlor is tlms obtained pure enough for identification. 



The writer prefers, after fractionating the colors into the main 

 groups {is just described, to try the bromin test (page 47). The be- 

 havior with acids Inis already been seen in the com'se of the separation, 

 and that with alkaU can be qui(;k!y ascertained. Ordinarily these will 

 indicate the fraction to contam but one coloring matter. This is 

 then dyed out from a portion of tlio solution, and its shade and reac- 

 tions on the fiber with reagents nrv, compared with standards or with 

 statements in the tables, in which, to faciUtate compaj'ison, the dyes 

 have been arranged in the; order of their solubility. Since the color 

 61147°— Bull. 448-17 3 



