In a 5-l. round-bottomed flask are placed 2 kg. (1587 cc., 21.7 moles) of glycerol and 700 g. (585 cc., 12.9 moles) of technical 85 per cent formic acid (Note 1). The flask is connected with a condenser set for downward distillation and the temperature within the reaction mixture is indicated by a thermometer, the bulb of which is immersed in the liquid. It is advisable to use as a receiver a distilling flask attached tightly to the lower end of the condenser. A tube is then run from the side arm of the distilling flask to a bottle of strong caustic soda to dissolve and decompose any acrolein which may be formed; a few pieces of clay plate are added to the distillation flask containing the reactants in order to prevent bumping, and it is then heated rapidly over a good ring burner (Note 2). The first runnings should come over within fifteen minutes and a temperature of 195° should be reached in thirty to forty-five minutes. Slow heating causes charring and formation of much acrolein, and thus gives a very low yield of allyl alcohol. The distillate collected up to the point where the thermometer registers 195° is saved separately (Note 3). Heating is continued until the temperature reaches 260°, the main reaction taking place between 225° and 235°. At 260°, when the heating should be stopped, a white smoke appears and decomposition apparently begins. The distillate (about 750 cc.) coming over between the thermometer readings between 195° and 260° is saved. For this operation about four hours are required.

The contents of the flask are now allowed to cool to a temperature between 100° and 125°, and 500 g. (9.2 moles) more of the technical 85 per cent formic acid is added. The distillation is then repeated in exactly the same manner as described above and 500 cc. of distillate is collected between the temperatures from 195–260°. The reaction mixture is allowed to cool again and a third portion of 500 g. of formic acid is added. This distillation yields not more than 350 cc. of the desired fraction, indicating that the glycerol is used up and that further addition of formic acid is unnecessary; moreover, the residue left behind is now small, amounting only to 100–200 cc. The three distillations require from one to one and one-half days.

The 195–260° fractions of the distillates are treated with potassium carbonate to salt out the allyl alcohol and to neutralize the little formic acid present. This allyl alcohol is then distilled and the fraction boiling up to about 103° is collected, or if a column is used, up to 98°. In this way, 845 g. of an allyl alcohol is obtained, which by bromine titration shows a purity of about 68–70 per cent (Note 4). This is equivalent to 570–590 g. of pure allyl alcohol (45–47 per cent of the theoretical amount) (Note 5).

The alcohol may be made practically anhydrous by refluxing with successive portions of fused potassium carbonate until no further action is observed. The carbonate will remain finely divided and will not become sticky when water is absent. A considerable amount of allyl alcohol is lost mechanically during the drying in this way, so that the potassium carbonate which is used here should be employed for the salting out of fresh portions of allyl alcohol in the first part of subsequent preparations. The allyl alcohol thus produced is dry enough for all practical purposes (98–99 per cent), and it is unnecessary to dry with lime or barium oxide as advised in the literature in order to remove all the water. The allyl alcohol obtained by this process boils at 94–97°.

The checkers have used the following method with success for obtaining a completely anhydrous product. The allyl alcohol obtained after one salting out with potassium carbonate is distilled as indicated above to give a product of about 70 per cent purity and then mixed with one-quarter of its volume of carbon tetrachloride (Note 6). The mixture is then slowly distilled from a round-bottomed flask fitted with a fractionating column at least 80 cm. long, collecting the low-boiling mixture of carbon tetrachloride, water, and allyl alcohol, drying it with potassium carbonate, and returning it to the flask (Note 7). This process is repeated several times, until only very little water passes over. The mixture is then completely distilled, taking the following cuts: Up to 80°, 80–90°, 90–95°, 95.5–97°, the last being alcohol which titrates 100 per cent pure. The lower fractions are dried separately and refractionated, thus giving an additional portion boiling at 95.5–97°.