Benedikt and Zsigmondy1 published in 1885 a method for the estimation of glycerine by its oxidation to oxalic acid by permanganate of potash, precipitating the oxalic acid with calcium acetate, and determining it volumetrically by titration with acid. This method is also claimed by Fox and Wanklyn.2 At the time of the publication of this method I made several trials of it on pure glycerine with very satisfactory results, and Allen3 has confirmed the accuracy attributed to it by

1 Chem. Ztg. 9, 975; Analyst 10, 206.

2Chem. News 53, 15. 3 Analyst 11, 52.

Benedikt and Zsigmondy, and used it on saponified fats. It has never been applied to wine or beer, so far as I know.

Legler1 has formulated a method, intended, as the author says in his paper, to supply the place of the method of the Berlin committee, and atone for its deficiencies. It depends on the oxidation of glycerine to carbonic acid by means of sulphuric acid and potassic bichromate.2 The estimation of organic bodies by the oxidation of their contained carbon has been proposed and carried out by Cross and Bevan3 who operate in a dry, and by Burghardt,4 who operates in a wet, way. The operation was performed by Legler in a Will's carbonic acid apparatus, as follows:

The air flask contains the glycerine mixed with a saturated solution of potassic dichromate; the other contains, as usual, strong sulphuric acid. After the apparatus has been weighed, a little air is drawn out which causes some of the acid to mix with the chromate. A regular evolution of carbonic acid soon sets in, but must be assisted towards the last by gently boiling. The flask containing the sulphuric acid must be kept cool. When no more gas bubbles are formed, the apparatus is cooled by partial immersion in cold water, and the remaining carbonic acid is expelled by a current of dry air. The apparatus is now reweighed and the loss represents carbonic acid. The following equation shows the action taking place :

3C3H803+7K2Cr207+28H.2S04=7K2S04+14Cr2(S04)3+9CO2+40H1O.

One part of glycerine therefore requires about 7.5 parts at K2Cr2O7 and 10 parts of H1SO4, but an excess of each is of course used.

The mode of procedure in operating on wine is as follows: The crude glycerine obtained from 100cc. of wine, after evaporation with 3cc. milk of lime and 2 grams of quartz, and extracting the mass with alcohol of 96 per cent., is, after weighing, diluted up to a definite bulk, and aliquot parts are taken for the ash and the oxidation process. A white wine, containing 8.54 per cent. alcohol and 2.07 per cent. solid matter, gave in 100cc. 1.4 grams crude glycerine, with .1278 grams ash, 25cc. of the glycerine diluted up to 50cc. yielding .725 gm. CO2 = 1.10 per cent. glycerine. A duplicate experiment gave 1.47 crude glycerine with .136 ash, .710 CO2 == .99 per cent. glycerine, the average thus being 1 per cent.5 and the relation between alcohol and glycerine as 100 : 11.7.

Estimation of glycerine in wine after it has been purposely added.

Three lots of l00cc. each of the same wine were mixed respectively with .125, .250, and .500 grams of glycerine, and analyzed as before. The results were as follows:

Allowing for the 1. per cent. of natural glycerine in the sample wo obtain .115, .251, and .492 per cent. of glycerine.

1 Rep. Anal. Chem. 6, 631; Analyst 12, 14. 2 Chem.News 53, 297; also 55, 2. 3 Chem. News 55, 34 ; see also 55, 46.

4 Ibid.

5 The average is really 1.05 per cent., which would make some difference in the figures which follow,

The same wine was also submitted to an analysis by the old process, somewhat higher results being obtained, from which Legler concludes that his method gives lower but more correct figures. He also obtained promising results from beer and sweet wines. This method was tried on some of the samples analyzed in the Chemical Division, the analytical work being done by Mr. Felix Lengfeld.

The apparatus for the estimation of the carbonic acid was a specially designed one used for most carbonic acid estimates in the laboratory, and provided for the absorption of the dried gas by soda lime. A trial was first made with pure glycerine, which gave very satisfactory results, as follows : Weight of pure glycerine taken, .3645; weight glycerine obtained, .3605; difference, .004.

When it was applied to the wines, however, it was found that the results obtained varied widely from those obtained from the same samples by the old method, and instead of the variation being constant, it was found that sometimes the results were higher, sometimes lower, than by the old method, the manner of manipulation making a very marked difference in the quantities obtained. Thus when just about sufficient lime was taken to combine with the sugar present and the mass not evaporated very closely, higher results were obtained by Legler's process than by the old method, while if a large excess of lime was added and the whole evaporated very nearly to dryness the results were lower.

Thus Nos. 4998 and 4999, treated with considerable lime and evaporated nearly to dryness, gave the following results:

While Nos. 4995, 5000, and 5002, treated with a smaller amount of lime, and not evaporated so closely, gave:

From these and from several other determinations, where the differences were still more marked, it was concluded that Legler's method as it stands now cannot be relied on to give constant results with wines, for though undoubtedly an accurate method of estimating pure glycerine, it is open to the same objection that applies to the method by oxidation with permanganate, viz, that other organic bodies are also oxidized, when present, and give too high results. In the extraction of the glycerine from wine by alcohol other organic bodies are undoubtedly taken up by it (the most of which are removed in the old method, by the purification with ether-alcohol) which make the results too high. The only way that results could be obtained as low as the old method gave was as indicated above, by long treatment in the bath with a large excess of lime, and in these cases there was undoubtedly a loss of glycerine by evaporation, as the following experiments show:

1. .3645 grams pure glycerine were mixed with a small quantity of lime, alcohol and water added, the whole evaporated nearly to dryness on the water bath, extracted with alcohol, and the glycerine determined by Legler's method in the residue from the alcoholic solution; result, glycerine .303, or a loss of .0615.

2. .3045 grams glycerine were taken and treated as above, except that a large amount of lime was added; result, .275; loss, .0895.

3. .3045 grams were taken and treated as before, except that the evaporation was carried to dryness, the dish being allowed to stand on the bath about half an hour longer than in the other experiments; result, .251; loss, .1135.

Hehner1 has applied Legler's method to the estimation of glycerine in fats, with some modifications based upon the very important fact he claims to have established in his work, that "dilute solutions of glycer-ine (.6 glycerine to 1,000 of fluid) do not, as is commonly supposed, volatilize in concentrating the fluid, be it on the water-bath or over the naked flame."

Other methods that have been published by Benseman,2 Amthor,3 and Medicus,4 are essentially slight modifications of the Berlin method, and can only be referred to here. Sulman and Berry,8 in their article on "The Examination of Commercial Glycerine," give a very thorough resume of various methods in use up to the date of the article for the estimation of glycerine.

The most recent method published is by Diez,6 which I have not had time to investigate closely. The method is quite a novel one, and differs from the previously described methods in that it separates out the glycerine as an insoluble compound, viz, as a tribenzoate of glycerine. This is accomplished by shaking the (.5 to 1.0 per cent.) solution of glycerine with benzoyl chloride after an addition of alkali. As applied to dry wines it is described by the author as follows:7 "20cc. are evaporated to a moderately dry condition after the addition of lime. The residue is exhausted with 20cc. of hot, 90 per cent., alcohol. After cooling 30cc. of water-free ether are added, and filtered after standing, the filter being washed with water-free alcohol ether (2:3). After the evaporation of the solvent the glycerine is dissolved in 10-20cc. of water.

1 Analyst 12, 44.

2Chem. Zeit. 10, 554.

3Rep. der, Anal. Chem., 1880, 12.

4 I bid, 1886, 1.

5 An.hst 11, l.'an.l 34.

6 Zeit. f. Physio. Chem. 11. 472. 7 Ibid., 480.

according to the quantity, and shaken up with 5cc. benzoyl chloride and 35cc. of 10 per cent soda solution for 10-15 minutes without cessation and with frequent cooling. The estergemenge or precipitate obtained, is collected upon a filter, washed and dried for 2 to 3 hours at 100° C.; 0.385 gram of the weighed precipitate corresponds to 0.1 gram glycerine." The objection to this process seems to be that it is not applicable directly to the wine or beer, but the glycerine must be separated out in a state of comparative purity before it can be converted into the benzoate, and there is still the liability to loss during the process of purification. The results given in the table are by the old method.