To 225 g. (0.87 mole) of crude 2,4,6-trinitrobenzoic acid (p. 543) (Note 1) and (Note 2) is added 2100 g. (1800 cc., 19.6 moles) of concentrated hydrochloric acid (sp. gr. 1.17) in a 5-l. flask provided with a reflux condenser, the inner tube of which is 15–20 mm. in diameter. To this mixture is added 830 g. (7.0 atoms) of granulated tin in small portions, the reaction being started after the addition of 30–40 g. of the tin by immersing the flask in a hot-water bath (Note 3). The rest of the tin is added through the condenser at such a rate that the reaction proceeds briskly. It is advisable to carry out this reaction under a hood since hydrogen chloride is likely to escape at the beginning of the reaction. The mixture is finally heated on a steam bath for one hour in order to complete the reaction, and is then filtered hot through glass wool in order to remove unreacted tin.

The filtrate is made up to exactly 2 l., and a 50-cc. sample is titrated with sodium hydroxide solution of known concentration (35–40 per cent in a suitable strength). The volume of alkali necessary to produce a slight permanent precipitate is noted (it usually requires 6–8 cc. of 40 per cent alkali). More alkali is run in until the liquid is neutral to litmus paper: this second value represents the acid combined with tin. To the filtrate in the 5-l. flask are added 1 kg. of cracked ice and then sufficient alkali to neutralize the free acid as well as 60 per cent of that combined with the tin (about 800 cc. of 40 per cent sodium hydroxide is required). The mixture is diluted, without filtering, to 6 l. and boiled in a 12-l. flask under a reflux condenser in an atmosphere of coal gas (Note 4) for twenty hours. The precipitate is then filtered off and washed very thoroughly with boiling water (2–3 l. of water); the combined filtrates are concentrated to 3 l., then rendered slightly acid to litmus with hydrochloric acid, and chilled to 0°, when crystals of phloroglucinol separate. This crop is filtered off; it should weigh 70–80 g. It is then dissolved in 700 cc. of hot water, filtered, and allowed to crystallize at 0°. In this way 65–75 g. (46–53 per cent of the theoretical amount) of almost colorless (Note 5) phloroglucinol dihydrate melting at 217–218° is obtained.