208 



not large enough to make a satisfactory separation, while the water-insoluble method 

 shows a much wider difference and gives the same information, but is not so valuable 

 as the amyl alcohol test, which makes the most complete separation of the two kinds 

 of coloring matter. In a study of the water-insoluble method it was found best to 

 evaporate just to dryness, and, further, that the manner of evaporation did not affect 

 the results. It also appears that the amount of sugar present as caramel does not 

 affect the solubility of the whisky coloring matter. 



DETERMINATION OF FUSL OIL. 



The determination of fusel oil or higher alcohol is undoubtedly one of the most 

 important ones made in the analysis of distilled spirits, giving more information as 

 to the methods of distillation in the manufacture of the spirit than any other factor. 

 When the examination of distilled spirits was begun, an extensive investigation was 

 made of the Roese method as given in the official methods of the Association of Offi- 

 cial Agricultural Chemists, and it was found that a great many difficulties were 

 encountered in employing the apparatus and method as there directed. This method, 

 depending as it does upon the relative solubilities of alcohol, chloroform, and fusel 

 oil in each other, requires that the conditions of temperature and concentration must 

 be very carefully controlled. The first difficulty encountered was the leaking of the 

 stopcocks in the apparatus adopted by the association, known as the Bromwell tube, 

 and after many experiments it appeared that this could not be overcome. The 

 chloroform solution would invariably leak through the ground-glass stoppers, so that 

 it became necessary to return to the older form of apparatus as designed by Roese, 

 which has no stopcock, but is extremely difficult to fill. With this form of appa- 

 ratus, however, somewhat satisfactory results were obtained. 



It was found also to be absolutely necessary that the apparatus be perfectly clean 

 and free from any oily material, and in order to insure this it was heated in a sulphuric- 

 acid-bichromate solution after almost every determination. Unless this is done 

 drops of water will stick to the sides of the chloroform bulb and increase the volume 

 of the chloroform and the amount of fusel oil shown. Also it is absolutely necessary 

 that during the whole determination the solutions and apparatus should be kept 

 exactly at 15 C., and to this end a large constant temperature bath was built deep 

 enough so that the tubes could be immersed completely and the shaking could be 

 carried on in the bath itself. It was found that if the tubes were removed from the 

 bath and shaken in the air the results were entirely inaccurate, a much larger blank 

 being obtained. This is easily explainable. If the temperature of the room is very 

 much above that of the bath, the shaking will raise the temperature of the solution 

 and change the relations between the solubilities of the various liquids in each other, 

 thus yielding results of little value. 



A regular procedure was adopted in regard to the shaking. The apparatus was 

 filled according to directions and immersed in the constant-temperature bath until 

 all the solutions had reached the same temperature. The tube was inverted in the 

 bath and shaken vigorously 150 times, then reversed and allowed to stand in the tank 

 until all the chloroform had settled back into the bulb, after which a reading was 

 made. 



By using this apparatus and carefully following these details fairly satisfactory 

 results were obtained, but at the same time the Allen-Marquardt method was tested 

 and found to be much more convenient and accurate. In a recent paper by Doctor 

 Dudley, of Vanderbilt University, on "The comparison of results obtained by the 

 Roese and the Allen-Marquardt methods, "& these same difficulties and errors in the 

 Roese method were noted, so that it seems advisable to abandon the old Roese method 

 and direct our attention to the Allen-Marquardt method, which apparently gives 

 much more satisfactory results. 



a Bui. 107, Revised, p. 97. 6 J. Amer. Chem. Soc., 1908, 30 : 1271. 



