precipitated, fractionally distilled under diminished pressure (about 19 mm.). A small quantity boils below 190°; it contains bromine, and is most probably a mixture of ethylic malonate and ethylic dibromomaleate. A considerable amount distils between 190° and 230°, and at 230-250° a thick oil passes over, whilst at the same time decomposition occurs. The fraction which boils between 190° and 230° partially solidifies to a crystalline mass, which is freed from adhering oil by the aid of a pump, and recrystallised from alcohol. In this way, long, colourless needles are obtained which melt at 76°. The identity of this compound with ethylic ethanetetracarboxylate, (COOC2H),CH CH(COOC2H), has been ascertained by comparison with a specimen prepared by Bischoff's method, and also by the following analyses : 2 2 0.2163 gave 0.4167 CO, and 0.1350 H2O. C=52·54; H=6·93. 0.2261 04375 CO,,, 01452 H,O. C=5277; H=713. C14H22O8 requires C=52.83; H=6.92 per cent. The properties of this compound agree with those of the substance which Pum obtained, and which he regarded as ethylic dimalonylmaleate. The fraction of high boiling point produced in the above reaction undoubtedly contains ethylic propenetetracarboxylate, which owes its formation to the union of the ethylic salts of malonic and acetylenedicarboxylic acids. Our attempts to obtain this compound in a pure state by redistillation in a vacuum have been unsuccessful, owing to partial decomposition, a fact also observed by Michael (J. pr. Chem., 1894, [ii], 49, 20), yet the analyses of the fractions boiling at (I) 240-243° and (II) 249-250°, respectively, under 19 mm. pressure, give results sufficiently close to those required by the formula COOC,H, CH:C(COOC,H,)•CH(COOC,H,), I. 0.2268 gave 0·4475 CO, and 0·1385 H2O. C=53·81; H=6·78. In conclusion, we express our thanks to the Government Grant Committee of the Royal Society for assistance in carrying out this investigation. GONVILLE AND CAIUS COLLEGE, CAMBRIDGE. XCIII-A Contribution to the Chemistry of the Mandelic Acids. By ALEX. MCKENZIE, M.A., D.Sc. THE following work was undertaken in order to obtain d- and l-mandelic acids for the preparation of the optically active phenylalkyloxyacetic acids recently described (Trans., 1899, 75, 753). Although recourse was eventually had to amygdalin as a convenient source of l-mandelic acid, the results obtained in a series of experiments on the resolution of the inactive acid into its optical isomerides are put on record, in view of the attention at present being given to the general subject.* Mandelic acid had previously been resolved into its active components by Lewkowitsch (Ber., 1883, 16, 1569), who isolated the d-acid, first by employing Penicillium glaucum, and later by means of a nucleus of cinchonine d-mandelate; for the preparation of the l-acid, he used Saccharomyces ellipsoideus. As I had not cinchonine d-mandelate at my disposal, the inactive acid was accordingly resolved by alkaloids. Preliminary experiments with cinchonine, cinchonidine, brucine, and morphine, in aqueous solutions, and with cinchonidine, in alcoholic and benzene solutions, showed that the difference in solubility between the alkaloidal d- and l-salts was not such as to afford a convenient method. The partial resolution with cinchonidine may be quoted, as it possibly illustrates Ladenburg's views regarding the resolution of an acid by alkaloids. Partial Resolution by Cinchonidine. 20 grams of mandelic acid were dissolved in 1200 c.c. of water, and the boiling solution neutralised with the calculated quantity of alkaloid. The crystalline salt (I), which separated on cooling, was collected, and from the filtrate, after a day's interval, a further crop (II) was removed. As the two crops were equally active, they were united and recrystallised several times, the progress of the separation being noted as follows. One gram of the air-dried salt was dissolved in the requisite amount of water; the cinchonidine which was precipitated by a definite volume of standard ammonia solution was collected and washed with water, the washings being added to the solution of the ammonium salt; the latter was next evaporated to less than the volume required to fill a 4 dcm. tube, and when just ammoniacal was made up to the volume necessary to fill the tube. The separation is represented by the fol * Rimbach's paper, "Ueber Spaltung und Eigenschaften der Mandelsäure ” (Ber. 1899, 32, 2385) appeared after this work had been sent in for publication. lowing scheme, the rotations being those observed for the ammonium The acid obtained from V by precipitation of the alkaloid with ammonia, acidification of the ammonium salt, and extraction with ether, melted at 113-117°, and had the specific rotation +49°, whilst the pure d-acid, described by Lewkowitsch (loc. cit.), had for similar concentration the value +156°. The positive rotations of the ammonium salts were probably lowered slightly by the presence of a little lævorotatory cinchonidine dissolved in the process of washing, but the values are comparable, as like conditions were observed in each determination. The rotations of the ammonium salt in IV and V are practically identical, and yet the acid from V had not nearly reached the full activity. It is therefore quite possible that in this case a partially racemic alkaloidal salt had been formed. Resolution by Quinidine. 100 grams of mandelic acid were dissolved in 4850 c.c. of water, and the hot solution neutralised by the calculated quantity of quinidine. On cooling, the mixture of the alkaloidal salts tended to separate as an oil, but crystallisation soon started in the form of needles. The following scheme indicates the method of separation : Crop 8, after removal of the alkaloid with ammonia, yielded 21 grams of an almost pure, active mandelic acid (m. p. 127–131°), from which, after two crystallisations from water, the pure l-acid was obtained, melting at 132°, and with the specific rotation - 156.3° for c= 3.093. Lewkowitsch gives the melting point as 132.8°, and the specific rotation for similar concentration as 156.4°. The acid obtained from crop, after one recrystallisation from water, melted at 122-126°, and, together with the lavorotatory aqueous mother liquors from the crystallisation of the pure acid, was worked up subsequently. The acid obtained from filtrate II was twice crystallised from water and yielded the d-acid, melting at 132-133° and with the specific rotation +158.2° for c= = 3.5187. Preparation of the Active Acids by means of Cinchonine. By the addition of a nucleus of cinchonine d-mandelate to an aqueous solution of mandelic acid which had been previously neutralised by cinchonine, Lewkowitsch obtained at once a crop of pure cinchonine d-mandelate, and from this the pure d-acid. On repetition of his method under varying conditions, an oil usually separated on cooling the solution, and I invariably obtained a mixture of the alkaloidal salts (or, possibly, a partially racemic salt). The cinchonine d-salt, used as a nucleus, was crystalline, and was prepared from the d-acid, obtained by the quinidine resolution just described. One experiment, in which the nucleus was added to the hot solution before the separation of oil and remained undissolved, yielded a salt which gave an acid melting at 115-125°, and containing, according to Lewkowitsch, approximately 40 per cent. of the active form. The success of this resolution doubtless depends very materially on the temperature at which crystallisation of the d-salt is induced. For the preparation of the active mandelic acids on a larger scale, 200 grams of inactive acid were dissolved in about 13 litres of boiling water and neutralised with the calculated quantity of cinchonine. Before any turbidity was evident, crystallisation was induced in the warm solution by a nucleus of cinchonine d-mandelate. The salt which separated yielded about 90 grams of active acid, which was recrystallised from water until the melting point was sharp. The pure acid thus obtained had the following rotation: l=2·2, c=3·9525, aoû=+13·80°, [a]»= +158·7°. The filtrate from which the cinchonine salt had been separated measured about 8 litres. The l-acid, obtained from this, was repeatedly crystallised from water until it melted sharply; it then had the following rotation: l=1, c = 3.8295, a2 = To obtain a further quantity of -- 6·12°, [a] = - 159.8°. d-acid, the mother liquors of the dextro-acid were again neutralised with cinchonine, and the treatment repeated as previously described. The mother liquors of the lævoacid were also neutralised with quinidine, and by help of a nucleus of quinidine l-mandelate yielded a further quantity of l-acid. The difference in solubility between the inactive and active acids in water, the inactive at the ordinary temperature being about twice as soluble as the active, was utilised for the isolation of the active acids in the preceding experiments. On repeated evaporation of aqueous solutions, however, it was found that the separation of the solid acid became more difficult from the formation of a slight amount of a smeary product, possibly an anhydride. The acids also are partially decomposed at 100°: for instance, 0·6518 gram of the inactive acid lost 0·1083 gram when heated for 41 hours at 100°, and 0·6082 gram of the d-acid, under similar conditions, lost 0.0830 gram. Benzaldehyde is slowly evolved, and the white acids become vitreous and yellowish with the formation of diphenylmaleic anhydride. From the pure active acids, the salts of various alkaloids were prepared by neutralisation with the calculated amount of alkaloid. In each case, a crop was withdrawn from the solution and air-dried on clay. The solubilities in water at 18-20° were determined by Victor Meyer's method, S representing the number of parts dissolved by 100 parts of water. The quinine salts are slightly acid to litmus, and it appears that their solutions cannot be evaporated without hydrolysis taking place, |