partial change of rotation of these sugars with temperature can be applied as a correction.

Gubbe's [4] comprehensive formulas for the specific rotation of invert sugar can be solved for the change of its rotation with change. of temperature or for the temperature of complete inactivation. This inactivation is due to the fact that since the rotatory power of levulose diminishes with increasing temperature, while that of dextrose remains constant, there is some temperature at which the two rotations are equal and opposite. Gubbe's formulas

[a]=-19.657 -0.0361C

[a] = [a] +0.3246 (t-20)-0.00021 (t-20)2

(53)

(54)

indicate that the specific rotation at 20° C is variable with concentration but that the temperature coefficient is independent of concentration. The temperature of inactivation therefore is not a constant but a function of concentration.

Browne [3] has calculated that the temperature of inactivation varies with concentration of invert sugar from 83.2° C for 2 g to 90.2° C for 60 g in 100 ml. For general purposes, 87° C is usually taken as the temperature of optical inactivity of invert sugar.

On the other hand, the temperature coefficient of the rotation of invert sugar is 0.0180° S for each gram in 100 ml, regardless of concentration. Hence invert sugar without any assumption of a temperature of inactivation can be determined by

P'-P 0.0180 (t'-t)

=grams in 100 ml,

(55)

in which P' is the saccharimetric reading at t' and P the reading at t in a 200-mm column. Obviously, any pair of temperatures sufficiently separated will serve for the determination.

The suggestion seems valid that the method of determination by temperature coefficient is more reliable than that of polarization at the temperature of inactivation. Not only is there considerable experimental difficulty in accurately maintaining a temperature of 87° C, but there remains the uncertainty that inactivity has been attained. The method of temperature coefficient is free from these uncertainties, since any pair of temperatures will serve, provided they are accurately observed.

After the temperature coefficient has been determined, the invert sugar is calculated by formula 55. If it is desired to determine the other constituent of the mixture, the rotation of the invert sugar at 20° C (P20) can be substituted into the low-temperature polarization by reference to table 76, p. 563, which is computed from the Gubbe specific rotation formula 53, in which C refers to the concentration of total solids in 100 ml. The third column gives the rotation which each gram of invert sugar contributes to the total rotation at various concentrations of sugar. Note that these values are expressed in saccharimeter degrees. P20 so found can be transformed to P, if necessary, by formula 55. The rotation of invert sugar is deducted algebraically from the observed polarization, leaving a remainder which represents the rotation of the second constituent. If there is an excess of dextrose, it can be calculated quantitatively

by dividing its rotation by that of 1 g of dextrose at the concentration of total sugars in the mixture. The rotation of dextrose can be selected from table 74, p. 562.

Frequently the second constituent of the mixture is commercial ~ glucose. This product as manufactured in this country is a liquid of density varying from 41° to 45° Baumé, and has a specific rotation varying from 100° to 125°. Obviously no exact determination is possible by means of polariscopic measurement, but if a specific rotation of say 108° is arbitrarily assumed, a measure of the constituent is obtained by dividing the observed rotation (corrected by deducting the rotation of the invert sugar) by 0.1600, the polarization of 1 g of the liquid product. The analyst should always state the specific rotation which is assumed for the purpose of calculation.

Any other probable specific rotation can be assumed for the purposes of this calculation, and the appropriate divisor can be found in table 28, which is taken from Browne's Handbook of Sugar Analysis [3].

TABLE 28.-Rotatory power of commercial glucose

(c) LEVULOSE BY POLARIZATION AT TWO TEMPERATURES

By the procedure outlined in the previous section, levulose can be determined by polarization at two temperatures. The change of rotation of 1 g of levulose per degree change of temperature should be exactly twice as great as that of invert sugar, or 0.036. Wiley [5] gives the value 0.0357. The average computed value of five previous investigations [3] is 0.0362, the difference between the extreme values being about 20 percent. On the other hand, Jackson and Mathews [6] in an extended investigation found experimentally between 20° and about 70° C a coefficient of 0.03441. The value was found to be independent of concentration between 3 and 18 g of levulose in 100 ml. There is thus an outstanding discrepancy of 4.5 percent between the older values of the temperature coefficient and the recent experimental determination of Jackson and Mathews. The lower value of the coefficient has been closely verified by Lothrop [7], who found in a limited number of experiments between 20° and 70° C a mean value of 0.0341. By application of the Jackson and Mathews coefficient, 0.03441, to the analysis of levulose in honey, Lothrop found a close agreement with the levulose percentage as determined by chemical methods.

This important coefficient requires further investigation. Not only is it divergent from Wiley's value, but it is inconsistent with the invert-sugar coefficient, 0.018, which should be exactly half that of levulose. It is apparent that the coefficient remains constant with varying concentration of sugar, but to what extent it is constant between different temperature intervals is at present undetermined.

Until further investigations are made, it is to be recommended that the temperature interval 20° to 70° C be employed with the coefficient 0.03441, that is, 1 g of levulose in 100 ml of solution in a 200-mm column diminishes 0.03441° S for each degree rise of temperature. The mean expansion coefficient between these temperature limits is, according to Jackson and Mathews, 0.00044. Thus if the higher temperature is exactly 70°, the observed polarization must be multiplied by 1.022 before subtracting from the polarization at 20° C. The decrease in the corrected polarization divided by 0.03441 yields the number of grams of levulose in 100 ml of solution.

(d) GALACTOSE BY MUCIC ACID PRECIPITATION

Galactose is oxidized by nitric acid to yield about 75 percent of mucic acid. Under closely specified conditions of analysis, the quantity of recovered mucic acid is reproducible and can be related empirically to the quantity of galactose in the sample. The method is applicable to free galactose or to the combined galactose in compound sugars or in galactans. Certain glycosides containing galactose, for example, saponins, yield insoluble products upon hydrolysis. Such glycosides must first be hydrolyzed with sulfuric acid (2 to 5 percent) and the insoluble material separated by filtration [8].

van der Haar [8] has given detailed specifications for the analytical procedure. Transfer the weighed sample containing galactose to a beaker (12 cm in height and about 60 mm in diameter) and add sufficient sucrose to increase the weight of total sugar to 1.000 g. Add 60 ml of nitric acid (sp gr 1.15 at 15° C) and place the beaker in an inclined position in a boiling-water bath and with repeated agitation allow it to remain until the weight of the contents has diminished to somewhat less than 20 g (that is, 19.8 to 20). Cool, and add water to make the weight exactly 20 g. Add 500 mg of pure, dry mucic acid and allow to stand for 48 hours at approximately 15° C, during which time stir occasionally. During the last few hours, adjust the temperature to exactly 15° C. Filter the precipitated mucic acid with suction on a weighed Gooch crucible prepared with asbestos which has previously been treated with nitric acid. Wash the precipitate four times with 5 ml of a solution of mucic acid saturated at 15° C and finally with 5 ml of water. Dry the precipitate at 100° C to constant weight. Deduct 500 mg from the weight of the precipitate and refer the result to column 3 of table 103, p. 608.

Acree [9] states that the oxidation of galactose by nitric acid is accelerated by the oxides of nitrogen, hence if the nitric acid is too pure. it is preferable to add a small quantity of nitrous acid or an alkali nitrite.

(e) DETERMINATION OF. MANNOSE AS PHENYLHYDRAZONE

While all reducing sugars are capable of forming hydrazones, the hydrazone of mannose is particularly insoluble and thus is suitable for its quantitative estimation. Bourquelot and Hérissey [10] prescribe the conditions of analysis. About 1 g of mannose dissolved in 16.6 ml of water is treated with a solution of 1.2 ml of phenylhydrazin and 1.2 ml of glacial acetic acid made up to 6 ml with water, and allowed to stand for 8 hours at a temperature not above 10° C. The hydrazone is collected on a Gooch crucible and washed with 15 ml

of ice water, 10 ml of absolute alcohol, and 10 ml of ether. The precipitate is dried in a vacuum over sulfuric acid. One gram of mannose yields theoretically 1.5 g of phenylhydrazone.

The hydrazone is soluble to the extent of 40 mg in 100 ml of solution and a small correction for this solubility increases the precision of analysis.

Pellet [11] has found the method suitable for the estimation of small amounts of mannose in cane molasses.

(f) DETERMINATION OF ARABINOSE AS DIPHENYLHYDRAZONE

Neuberg and Wohlgemuth [12] have made use of the high insolubility of arabinose diphenylhydrazone for estimating arabinose in the presence of other monosaccharides. Mannose or fucose in excessive quantities are apparently the only sugars which interfere with the selectivity of the analysis. The authors illustrate the method by the following example.

A solution (100 ml) containing dextrose, fructose, xylose, glucuronic acid, and 1.0066 g of arabinose was evaporated to 30 ml. The resulting solution was heated on a water bath for 1⁄2 hour with 6 g of diphenylhydrazine in 50 ml of 96-percent alcohol with additions of very dilute alcohol or preferably with a reflux condenser. The solution was cooled and allowed to stand for 24 hours. The precipitated hydrazone was collected on a Gooch crucible, washed with 50 ml of 50-percent alcohol, dried, and weighed. Yield, 2.1143 g of hydrazone, equivalent to 1.0035 g of arabinose. Factor, 0. 4747.

(g) DETERMINATION OF URONIC ACIDS

The uronic acids, glucuronic and galacturonic, are widely distributed in both plants and animals. They play an important role in the carbohydrate metabolism of the cell wall. Dickson, Otterson, and Link [13] have found that free glucuronic acid is present within the cell of corn seedlings and that a polymerized glucuronic acid sometimes associated with the cellulose, comprises part of the pectinaceous substance of the cell and cell wall. Nanji, Patin, and Ling [14] found that purified pectin preparations contained from 70 to 73 percent of uronic acid anhydride. Browne and Phillips [15] showed that uronic acids comprised about 3 percent of sugar-cane bagasse and that sugar-cane juice contained from 0.1 to 0.6 percent (based on ash-free solids) of uronic acids, the variations depending upon the methods of maceration. In cane molasses the uronic acids were found concentrated to an average of about 2 percent. These authors believed that the uronic acids are derived from pectins which are extracted with the juice.

When a uronic acid is heated with hydrochloric acid, decarboxylation occurs with the formation of furfural and carbon dioxide according to the equation.

The yield of furfural is less than the theoretical, while that of carbon dioxide is quantitative. In the absence of other reactions yielding carbon dioxide, a measure of the gas evolved serves for the quantitative determination of uronic acid.

Whistler, Martin, and Harris [16] in a study of the determination of uronic acids in cellulosic materials, found that under the drastic conditions employed in the analysis, carbohydrates free from uronic acids were slowly but regularly decomposed with the formation of carbon dioxide, necessitating the application of a correction.

The method of determination was originally devised by Lefèvre and Tollens [17]. Dickson, Otterson, and Link [13] further elaborated the method, measuring the carbon dioxide by absorption in barium hydroxide and titration of unreacted alkali.

The details of the method described here are those of Whistler, Martin, and Harris [16], who adapted the procedure specifically to the determination of uronic-acid groups in cellulosic materials. Their procedure can, however, be used without modification for any other material by selecting a weight of sample which will yield 30 to 40 mg of carbon dioxide.

FIGURE 39 (a).-Apparatus for determination of the rate of evolution of carbon dioxide from uronic acids or materials containing uronic acids during treatment with hydrochloric acid.

The apparatus is not drawn to scale. See original article for dimensions.

The apparatus is shown in figure 39 (a). Nitrogen, which is used as the carrier gas for the evolved carbon dioxide, enters the apparatus through an empty safety bottle, A. It next passes through an alkaline solution of pyrogallol, B. The inlet tube in this bottle is drawn out to a small orifice, which produces very fine bubbles. From B the gas passes through two absorption towers, C, filled with soda lime, into a second safety bottle, D, which is provided with a mercury manometer, E, and then enters the reaction flask, F, by way of a side arm whose outlet is 10 to 15 mm above the surface of the liquid in the flask. The size of the flask depends upon the type of material and the size of the sample to be analyzed. In most experiments a 500-ml reaction flask is suitable. From the reaction flask the gas passes through a 40-cm reflux condenser, G, and into a bubbling tower, H, containing approximately 60 ml of concentrated sulfuric acid. The sulfuric acid serves to remove interfering decomposition products which are carried over from the reaction flask. The gas next passes through the U-tube, I, which is filled with anhydrous copper sulfate to remove chlorine or hydrogen sulfide, then through the tube, J,