bluish-black color with impure ferrous sulphate and a dark color with ferric chloride. It reduced ammoniacal silver nitrate and Fehling solution. These experiments indicated the presence of a tannin compound. An alcoholic solution of the resin gave the same color reactions with iron salts as did the potassium salt. To determine which one of the tannin compounds was present was a matter of some difficulty since the di- and tri-hydroxybenzoic acids have in general the same color reactions. The presence of other plant substances in the solution also interferes with the color tests, and finally, a substance which gives a blue color with iron salts and one giving a green color may be found together in the same plant.[19] Further tests with a solution of the resin in dilute alcohol, and with a water solution of the acid precipitated by adding sulphuric acid to a solution of the resin in potassium hydroxide, led to the conclusion that the acid is gallic acid. These tests were the following:

(1) Boiling with an excess of potassium hydroxide gave a black substance (tauromelanic acid).

(2) The acid was not precipitated by gelatin.

(3) On addition of potassium cyanide a transitory red color appeared which reappeared on shaking with air.

Gallic acid is distinguished from tannic acid by tests (2) and (3). At later stages in the work the potassium, barium, and sodium salts of gallic acid were obtained, and finally the pure acid was made by decomposing the sodium salt with sulphuric acid and crystallizing from water. A portion of the acid so obtained was further purified by dissolving in absolute alcohol and pouring into absolute ether.[20] The melting point behavior of the acid corresponds with that of gallic acid; it melted with decomposition at about 230°. For further identification, some of the acid was converted into an ester by the following process: it was dissolved in 80 per cent. alcohol, hydrochloric acid gas was passed in, and the solution was heated an hour on the water bath. It was then evaporated to a small bulk, neutralized with barium carbonate and extracted with ether. The ether, on evaporation, left the ester which was crystallized from water and dried in a desiccator over sulphuric acid. The anhydrous ester agreed in melting point (156° to 159°) and other properties with the ester of gallic acid described by Grimaux.[21] For the sake of comparison, an ester was made from gallic acid obtained from another source and the two agreed in properties. A mixture of the two esters melted within the limits given for the ester of gallic acid.

While the tests leading to the identification of gallic acid were being made, another series of experiments was in progress. Eleven and one-half grams of the resin obtained from lead precipitate A by decomposition with hydrogen sulphide were treated with 0.1 n. potassium hydroxide added from a burette until the acid was exactly neutralized. All went into solution. On shaking with ether, some of the potassium salt separated out and was saved for examination. The solution became brown on exposure to air and got darker as the work proceeded. The acid in solution as a potassium salt was precipitated out in four fractions by adding for each fraction one-fourth the amount of 0.1 n. sulphuric acid required to neutralize the potassium hydroxide used. The precipitates were filtered off and examined. The first was small in amount, gummy and hard to filter. The solution was shaken with ether after each precipitate had been filtered off. The succeeding precipitates were in better condition, but were not pure. All appeared to be impure gallic acid which had become brown by absorption of oxygen. They were saved, however, to be tested for poison. After the last fraction had separated, the filtrate was shaken several times with ether and saved for further examination, which will be described under "Rhamnose." This filtrate is designated as B.

At this stage of the work a portion of the resin obtained from lead precipitate A was tested and found to be not poisonous as already mentioned. By this test, all the substances contained in the lead precipitate A after its extraction with ether in the Soxhlet apparatus, were eliminated from the possible poisonous substances. The poison must therefore have been extracted by the ether.

A fresh portion of the original poisonous material was treated with 50 per cent. alcohol and filtered from insoluble tar. The filtrate was precipitated in six fractions by lead acetate. The last fractions were lighter in color and apparently much purer than the first. The sixth lead precipitate was decomposed by hydrogen sulphide, the light-yellow water solution was tested and found to be not poisonous. It gave the characteristic reactions for gallic acid. The poison, if precipitated at all by lead acetate, must have gone down in one of the preceding fractions. Later experiments showed that it is brought down partly mechanically and partly as a lead compound in the first precipitates.

FISETIN.

Having identified gallic acid, and not finding any other phenol derivative in the lead precipitate, some of the original material was extracted with hot water to remove gallic acid and filtered from tar while hot. The filtrate had a deep yellow color. On cooling over night, an olive green precipitate separated out which was dried and found to be a light powder. It was practically insoluble in cold water, soluble with great difficulty in boiling water from which it separated in yellow flakes, slightly soluble in ether and in acetic acid, but readily soluble in alcohol. The solutions were not acid to litmus, gave a dark color with ferric chloride, an orange-red precipitate with lead acetate which was easily soluble in acetic acid, and an orange-yellow precipitate with stannous chloride. These properties and reactions indicated that the substance was the dye-stuff fisetin and that it occurs in the free state in this plant though it is usually found as a glucoside of fisetin combined with tannic acid. A compound of this kind was found in Rhus cotinus and named "fustin-tannide" by Schmid[22]. He showed that the fustin-tannide could be decomposed by acetic acid into tannic acid and a glucoside, fustin C₄₆H₄₂O₂₁. Fustin, on heating with dilute sulphuric acid, gave fisetin and a sugar supposed to be rhamnose. Fisetin was also found as a glucoside compound in Rhus rhodanthema by Perkin.[23]

The yellow substance which separated from the boiling water solution was further purified by dissolving in a small quantity of hot alcohol and adding hot water. On cooling, the yellow substance separated out in a flocculent condition. Examined under the microscope, the flakes appeared to be made up of masses of fine crystals.

An alcoholic solution of the substance gave a black color with ammonia which became red on addition of more ammonia. Concentrated acids intensified the yellow