According to Nussbaum,2 a mixture of equal volumes of absolute alcohol and light petroleum is homogeneous when heated slightly, but becomes turbid when cooled. The point at which the turbidity appears is sharply denned, but is raised by about 16° when the alcohol contains 1 per cent. of water.

1 Ann. Chun. Anal., 1908, 13, 4.

2 J. Pharm. Chim., 1917, 15, 230; J. Chem. Soc. (Abst.), 1917, 112, ii, 215.

The presence of water in alcohol may therefore be ascertained, and its quantity estimated, by determining the temperature at which the alcohol gives a turbidity when mixed with an equal volume of light petroleum. It is necessary, however, to make preliminary tests with light petroleum and quantities of alcohol containing definite amounts of water, because the point at which the turbidity is observed depends on the kind of light petroleum used.

The same principle - determining the critical temperature of solution - has also been applied by V. Rodt for the estimation of very small quantities of water in alcohol.1 The solvent employed is a mixture of transformer oil and commercial paraffin oil. To standardise it, perfectly anhydrous alcohol must be prepared, which is done as follows. About 2 litres of 99 per cent. alcohol are placed in a large flask with 300 grams of quicklime, and the flask is suspended in a water-bath which can be quickly removed. The alcohol is gently boiled for some days under a reflux condenser protected by a phosphorus pentoxide guard-tube. The condenser is then reversed and arranged to deliver the distillate into a wide, corked burette, also guarded with moisture-absorption tubes. Successive portions of alcohol are distilled over into the burette and tested as follows with the solvent until a constant minimum value is obtained.

For the test, all the apparatus must be dried in an oven at 150°, and every precaution taken against access of atmospheric moisture.

An Erlenmeyer flask of about 50 c.c. capacity fitted with a rubber stopper is accurately tared, a few drops of water are introduced rapidly from a pipette, and the weight noted. Then 15 to 25 c.c. of alcohol are drawn from the distillation burette and the flask is weighed again, so that the gravimetric composition of the aqueous alcohol is accurately known. The rubber stopper is replaced by another, which has also been dried in the oven, and has two holes; through one of these passes a standardised thermometer, reading to 02°, and through the other a short glass rod. The glass rod is withdrawn, the point of a burette being inserted through the hole, and a measured volume of the solvent is run in, calculated from its specific gravity to amount to 12/13 by weight of the alcohol employed. The measurement should be accurate to 0.1 c.c. The rod is then replaced; the contents of the flask are warmed with gentle rotation until the liquid is perfectly clear, and rotation is continued while cooling, until a sudden sharp clouding of the solution appears. The temperature at which this occurs is the critical temperature. It may be noted with an accuracy of 0.2°, which corresponds with 0.01 per cent. of water in the alcohol.

1 Mitt. K. Materialpruf., 1915, 33, 426; Analyst, 1916, 41, 316.

For a given oil, a standardisation curve is plotted, using different proportions of water to alcohol. This curve may be used subsequently for rapid and simple estimations of water in strong alcohol according to the foregoing procedure, mutatis mutandis. The rise of critical temperature is almost a straight line function of the percentage of water within the limits of 0 to 2 5 per cent.

M. Jones and A. Lapworth have used a-bromonaphthalene for the determination of water in moist alcohol.1 Weighed quantities of the alcohol and bromonaphthalene are mixed, and the solution is cooled down slowly until clouding appears. An opalescence occurs just above the critical point, but this is easily distinguished from the true critical point, when the liquid suddenly becomes opaque. The temperature at which this occurs is noted: it varies with the percentage of water in the alcohol if the concentration of the bromonaphthalene is constant, and also with the concentration when this varies. A table of values has been constructed, showing the percentage of water corresponding with various concentrations and temperatures: for this and details of manipulation, the original paper should be consulted.

Ethyl alcohol in the presence of methyl alcohol.

- Where a considerable proportion of ethyl alcohol is mixed with methyl alcohol, the refractometer, as already explained, will both show the presence and indicate approximately the proportion of the ethyl alcohol. Or the proportions of the two may be deduced from the density of the mixed iodides (Bardy and Meker's method); or the methyl alcohol may be estimated separately by one of the methods described earlier, and the ethyl alcohol taken by difference.

But when only a small quantity of ethyl alcohol is contained in a large volume of methyl alcohol, the problem of its detection and estimation becomes more difficult. With such proportions as one or two per cent. of ethyl alcohol probably Berthelot's well-known method is the best. It depends upon the conversion of the ethyl alcohol into ethylene by means of sulphuric acid, and the absorption of the ethylene in bromine with the formation of ethylene dibromide, which can be separated and measured. The details of this process as used officially in France are given by M. Louis Calvet2 as follows.

Into a flask of about 2 litres capacity fitted with a safety-funnel and connected with a wash bottle containing water, are poured

1 Trans. Chem. Soc, 1914, 105, 108C.

2 "Alcools," p. 154.

350 c.c. of sulphuric acid at 66° Be. To this is added carefully 50 c.c. of the methyl alcohol, avoiding too much rise of temperature. The reaction is allowed to proceed during half an hour, after which the flask is heated, at first gently, and then more rapidly and strongly. The vapours given off are passed through the wash bottle to remove sulphur dioxide, and collected in a bell-jar of 2 litres capacity, over water. When the jar is full, which will be in the course of about an hour and a half, the gases are passed through another-wash-bottle containing a solution of potassium hydroxide, and then through an absorption bulb containing bromine covered with a little water. To the outlet of the bulb a washing jar containing caustic potash solution is attached to absorb any bromine vapour carried over.

When the operation is finished, the contents of the absorption bulb are poured into dilute alkali solution, which absorbs the excess of bromine, leaving the ethylene dibromide undissolved, in the form of oily droplets. The volume of the dibromide is measured; each 01 c.c. corresponds with about 1 per cent. of ethyl alcohol in the liquid taken for analysis.

Deniges has described a method for detecting ethyl alcohol in the presence of methyl alcohol, which can also be used for estimating the proportion. The process is based upon the fact that ethyl alcohol is readily oxidised to acetaldehyde by means of bromine water, whereas methyl alcohol under similar conditions is not appreciably affected.1

A mixture of 0.2 c.c. of the alcohol to be tested and 5 c.c. of bromine water (0.3 c.c. of Br in 50 c.c. of water) is heated in a boiling water-bath for a period of five to six minutes, unless the colour is discharged sooner. The mixture is then cooled, and if still coloured, a solution of sodium sulphite (commercial, 36 to 40° Be., diluted to one-fifth strength) is added drop by drop until the colour just disappears. Then 5 c.c. of fuchsin-bisulphite solution are added, and the mixture allowed to stand for five to eight minutes. In the presence of ethyl alcohol a red or reddish-violet colour is developed, the intensity of which is proportional to the amount of ethyl alcohol present. One per cent. of ethyl alcohol can thus be detected, and the method can be made quantitative by comparing the colour with that obtained from standard solutions containing known quantities of ethyl and methyl alcohols.

The fuchsin-bisulphite solution is best prepared by adding 10 c.c. of sodium bisulphite solution (30° Be.) to 1 litre of an aqueous solution of fuchsin (0.1 per cent.), and when the colour has become faint, mixing in 10 c.c. of concentrated hydrochloric acid solution. With very small quantities a blank experiment on pure methyl alcohol is made, for comparison.

1 Bull. Soc. chim., 1910, 7, 951.

In the absence of methyl alcohol, or if only a small proportion is present, ethyl alcohol is detected by adding an equal volume of methyl alcohol to the sample under examination, and applying the test as described. In examining very dilute solutions (0.5 to 3 0 per cent.), the mixture to be heated is made by adding 0 2 c.c. of methyl alcohol and 0.03 c.c. of bromine to 5 c.c. of the alcohol for testing.