method and is subject to obvious variations with respect to details. Transfer to the precipitation vessel about one-tenth of the total quantity of milk-of-lime suspension and add one-tenth of the total volume of sugar sirup, allowing the latter to run in at a slow regular rate. It is advisable to examine the precipitate under the microscope from time to time and ascertain that the calcium levulate has formed discrete crystals. When properly prepared, they take the form of elongated prisms. Add successively the remaining portions of milk of lime, each being followed by the respective portion of sirup. Under favorable conditions the levulate crystals will be 0.1 to 0.2 mm in length and can be readily and rapidly separated from the waste water by filtration methods.

The yield of levulose by this process is dependent upon many circumstances. In order to obtain high yields, the temperature of precipitation must be in the neighborhood of 0° C. Increased temperature causes a considerable diminution in the yield of levulose. Thus, in a series of experiments on hydrolyzed artichoke juices, Jackson and Mathews [13] obtained a recovery of 82.3 percent at 3.2° C. 76.2 at 10.7° C, and but 64.0 at 16.0° C.

Typical quantitative data on the precipitation of calcium levulate from various juices are shown in table 46.

It will be observed that the "Total recovered" of each product (lines 16, 22, and 28) approximates the amount taken in the reaction mixture (lines 12, 18, and 24). The yield of levulose (line 33) varies inversely as the solubility of calcium levulate in the juice. The solubility of calcium levulate is greatly increased in the presence of nonlevulose sugars. Thus, in experiment 1, the levulate in the waste water, expressed as levulose (line 13), is held in solution largely by the dextrose. In the remaining experiments, it is dissolved both by dextrose and by other nonlevulose sugars.

Dextrose alone at 0° C holds in solution about one-fifth of its weight of levulose (in the form of levulate). Other nonlevulose sugars are derived from inulin, which upon acid hydrolysis yields, in addition to about 92 percent of levulose, 3 percent of dextrose and about 5 percent of a group of nonreducing difructose anhydrides [14]. It is due mainly to these latter substances that the waste water (line 13) from the plant juices contains so much more levulose than that from the mixture of pure sugars. Other nonlevulose sugars may be derived from too prolonged hydrolysis of the juices, which causes the formation of the levulosans described by Pictet and Chavan [15], or from a too alkaline medium for defecation, which may cause the Lobry de Bruyn-van Ekenstein transformation.

Table 46 shows in general the compositions of products and byproducts which would probably be found in a levulose plant of any magnitude.

The levulate cake, which contains in average about 22 percent of levulose and 10 percent of calcium oxide, is carbonated by adding it in small portions to a violently agitated quantity of cold water into which carbon dioxide is passed, preferably by some device such as the Doherty carbonator [16] which disperses the gas into a fine state of division. Carbonation is continued until the solution is slightly acid to phenolphthalein. The mixture at this point necessarily contains a considerable quantity of calcium bicarbonate which requires neutralization, since, if carried into the filtrate from the calcium carbonate, it would be decomposed during the subsequent evaporation by volatilization of carbon dioxide. When calcium bicarbonate is decomposed in this manner, it imparts a pH of 8 to 9 to the levulose solution. This alkalinity at the temperature of evaporation impairs the sugar seriously.

The most satisfactory method of neutralizing calcium bicarbonate is to add a thin suspension of milk of lime or calcium levulate until a minimum electrical conductivity is reached. The calcium carbonate, which at the low temperature sometimes crystallizes with six molecules of water of crystallization, is filtered and washed. The filtrate is now slightly alkaline, being saturated with calcium (and if present, magnesium) carbonate. It is again adjusted to a minimum electrical conductivity by addition of dilute oxalic acid.

The filtrate is evaporated at reduced pressure to a sirup. At some point during the evaporation there usually occurs a separation of inorganic salts, which are removed by filtration with 1 or 2 percent of active carbon. The resulting solution should have a levulose purity in excess of 99 percent, being contaminated mainly by inorganic salts in which magnesia usually predominates.

(5) CRYSTALLIZATION OF LEVULOSE.--Crystallization of the sugar can be conducted in aqueous or in aqueous alcoholic solution. For laboratory preparations, aqueous alcohol is the more convenient solvent. The solution is evaporated to a sirup and crystallized, after seeding, under the conditions illustrated in table 47. These conditions of course may be considerably varied.

Crystallization is accomplished in a crystallizer or in a tumbling machine or any apparatus which keeps the crystallizing mass in slow continuous motion. Frequently the addition of alcohol causes a separation into two liquid layers which, however, coalesce as crystallization progresses. The crystals are separated on a Büchner filter or by centrifugal drainage and thoroughly washed with 95-percent alcohol.

For the preparation of levulose of highest purity the substance can be recrystallized in the same manner. It sometimes occurs that the inorganic impurities persist. These can be removed by determining the ash content and, considering it as calcium carbonate, adding the stoichiometric equivalent of nitric acid. In this case the final washing with alcohol must be very thorough.

Levulose can also be crystallized from aqueous solution in spite of its high solubility in water. The solution is evaporated at diminished pressure to about 90 percent of solids, placed in a crystallizer at 50° to 55° C, seeded with crystals, and allowed to crystallize in slow motion by letting the temperature drop very gradually over a period of 1 to 3 or 4 days, depending upon the purity of the solution and the rate of growth of the crystals.

4. REFERENCES

[1] F. J. Bates and R. F. Jackson, Bul. BS 13, 633 (1916) S293.

[2] H. F. Bauer, Orig. Comm. 8th Int. Cong. App. Chem. 13, p. 21 (1912). [3] E. O. von Lippmann, Die Chemie der Zuckerarten 2, 1051 (F. Vieweg & Sohn Braunschweig, 1904).

[4] This Circular (p. 768).

[5] F. J. Bates and R. F. Jackson, Bul. BS 13, 67 (1916) S268.

[6] F. L. Dunlap, J. Am. Chem. Soc. 28, 397 (1906).

[7] E. Berner, Ber. deut. chem. Ges. 64, 842 (1931).

[8] B. Tollens V. H. Elsner, Kurzes Handbuch der Kohlenhydrate (J. A. Barth, Leipzig, 1935).

[9] J. R. Katz and A. Weidinger, Rec. trav. chim. 50, 1133 (1931). J. R. Katz and J. C. Derksen, Rec. trav. chim. 50, 248 (1931).

[10] H. Pringsheim, Ber. deut. chem. Ges. 62, 2378 (1929); 63, 2636 (1930). H. Vogel, Ber. deut. chem. Ges. 62, 2980 (1929).

[11] H. D. K. Drew and W. N. Haworth, J. Chem. Soc. 1928, 2690.

[12] R. F. Jackson, C. G. Silsbee, and M. J. Proffitt, BS Sci. Pap. 20, 587 (1926) $519.

R. F.

[13] R. F. Jackson and J. A. Mathews, J. Research NBS 15, 341 (1935) RP832. [14] R. F. Jackson and S. M. Goergen, BS J. Research 3, 27 (1929) RP79. Jackson and E. J. McDonald, BS J. Research 6, 709 (1931) RP299. [15] A. Pictet and J. Chavan, Helv. Chim. Acta 9, 809 (1926). [16] E. P. Clark, BS Sci. Pap. 17, 608 (1922) S432.