The indications furnished by the above test are rendered valueless, if the wine has been artificially colored by the addition of the coloring matter of grape-skins; but the execution of this fraud would require some knowledge of chemistry, and fortunately adulterators, as a class, are deficient in this branch of science. Another method for detecting the addition of water is based upon the fact that fermented liquors do not contain air in solution, but only carbonic acid; whereas, water dissolves oxygen and nitrogen. It is executed as follows: The wine to be tested is placed in a flask, the delivery-tube of which is also filled, and heated; the evolved gas being collected in a tube filled with mercury. In case the wine is pure, the disengaged gas will be completely absorbed by potassa; if, on the other hand, water has been added, an unabsorbed residue, consisting of oxygen and nitrogen, will remain. This test is useless in case water, through which a current of carbonic acid gas has been passed for a considerable time, has been employed. Under these circumstances, however, the presence of the gas would probably be detected by the taste of the wine, as well as by the estimation just mentioned, since the sample would invariably contain a larger proportion of the gas than the standard with which it is compared; indeed, it would be almost impossible to prepare a solution which contained exactly the proportion of carbonic acid ordinarily present in wine. It remains to mention the methods employed in determining the amount of alcohol and cream of tartar contained in wine. The alcometrical method usually employed is based upon the difference in density possessed by pure alcohol and by mixtures of alcohol and water. _Gay-Lussac_ has proposed an areometer (_alcoholmeter_), provided with a scale which directly indicates the proportion of alcohol contained in a mixture. As the indications furnished by this instrument vary with the temperature, and the scale is constructed on the basis of a temperature of 15°, a correction of the results obtained is necessary if the determination is made at other temperatures. Gay-Lussac has compiled a table which indicates at once the required correction; the following formula can also be used: _x = c ± 0.4 t_, where _x_ is the quantity of alcohol present in the sample; _c_ the degree indicated by the alcoholmeter, and _t_ the number of degrees differing from the temperature of 15°: the second member of the formula is subtracted from, or added to the first, as the temperature at which the estimation is made is greater or less than 15°.[W] [W] Tralles alcoholmeter is almost exclusively employed in this country.--_Trans._