From 1 to 1.5 grammes of the glycerine is weighed into a conical flask of about 150 c.c. capacity, 7 or 8 c.c. of acetic anhydride added, together with about 3 grammes of anhydrous sodium acetate, and the whole boiled on a sand-bath under a reflux condenser for one to one and a half hours, after which it is allowed to cool, 50 c.c. water added, and the ester dissolved by shaking, and gently warming, the reflux condenser still being attached as the acetin is very volatile. The solution is then filtered from a white flocculent precipitate, which contains most of the impurities, into a larger conical flask, of some 500-600 c.c. capacity, and after cooling, rendered just neutral to phenol-phthalein by means of N/2 caustic soda solution, the exact point being reached when the solution acquires a reddish-yellow tint; 25 c.c. of a strong caustic soda solution is then added, and the liquid boiled for about fifteen minutes, the excess of alkali being titrated after cooling, with N/1 or N/2 hydrochloric acid. A blank experiment is carried out simultaneously, with another 25 c.c. of the soda solution, and the difference in the amounts of acid required by the two, furnishes a measure of the alkali required to saponify the acetin formed, and hence the amount of glycerol in the crude glycerine may be calculated. _Example._--1.4367 grammes crude glycerine, after treatment with acetic anhydride, and neutralising, was saponified with 25 c.c. of a 10 per cent. caustic soda solution. The blank experiment required 111.05 c.c. N/1 hydrochloric acid. Flask containing acetin " 75.3 c.c. " " ----- 35.75 c.c. " " Hence, the acetin formed from the glycerol present in 1.4367 grammes of the crude glycerine required 35.75 c.c. N/1 caustic alkali for its saponification, so that the percentage of glycerol may be calculated from the following formula:-- 35.75 x 0.03067 x 100 Per cent. glycerol = --------------------- = 76.3. 1.4367 _Bichromate Method._--This process was originally devised by Hehner (_Journ. Soc. Chem. Ind._, 1889, 4-9), but the modification suggested by Richardson and Jaffe (_ibid._, 1898, 330) is preferred by the authors, and has been practised by them for several years with perfectly satisfactory results.