to table 97, p. 602, to ascertain the corresponding weight of invert sugar in the sample.

Standardization of thiosulfate.-Dissolve 25.5 g of Na2S2O3.5H2O and 0.2 g of Na2CO, and make to 1 liter. Standardize with pure K2Cr2O7.

Weigh accurately about 0.22 g of K2Cr2O, and dissolve in 450 ml of water in a 750-ml Erlenmeyer flask. Add successively 15 ml of KI solution (20 g in 100 ml) and 15 ml of concentrated HCI. Titrate the liberated iodine with thiosulfate, using 3 ml of a 1-percent starch. solution as indicator.

(a) SOMOGYI MODIFICATION OF SHAFFER AND HARTMANN MICROMETHOD FOR DEXTROSE

Shaffer and Hartmann [44] elaborated in detail a method for the determination of sugar in blood, which was applicable to the analysis. of samples containing from 0.07 to 2.2 mg of reducing sugar. The method was subsequently applied to other materials than blood and indeed has been found generally applicable to any material containing reducing sugar. It is particularly serviceable in instances in which it is necessary to conserve the supply of material.

Shaffer and Hartmann copper reagent was prepared by mixing copper sulfate, tartaric acid, and sodium carbonate, the last two substances reacting and releasing carbon dioxide. The solution also contained potassium iodate and potassium iodide in such concentration that it was 0.02 N with respect to iodine which was released upon acidification.

Somogyi [8], in a study of the method, found that the amount of copper reduced was in a high degree dependent upon the alkalinity of the solution. Within a narrow range, in which the ratio of Na2CO3 to NaHCO3 lay between 7:13 and 2:3 and the pH varied from 9.40 to 9.55, the highest copper values were obtained and the ratio of copper to sugar remained constant. In the Shaffer and Hartmann solution the ratio of carbonate to the bicarbonate formed by the reaction of tartaric acid with sodium carbonate was variable and depended upon the variable amount of carbon dioxide lost. Somogyi therefore modified the Shaffer and Hartmann solution.

Dissolve the Rochelle salt, sodium carbonate, and sodium bicarbonate in about 500 ml of water, and into this pour with stirring the copper sulfate dissolved in about 100 ml of water; then add the solution of the other constituents and dilute to 1 liter. (Only the potassium iodate need be weighed accurately.)

Measure 5 ml of the reagent into a large test tube (250 by 25 mm) and add 5 ml of the sugar solution containing not less than 0.1 mg and not more than 2.0 mg of dextrose. Mix by gentle shaking, cover the tube with a small funnel, bottle cap, or glass bulb, and keep it in a boiling-water bath for 15 minutes. Cool by placing in a shallow dish of water until the temperature is lowered to 35° or 40° C. Add with agitation 1 ml of 5 N H2SO, (or its equivalent) and see that all Cu2O is promptly dissolved. After about 2 minutes, titrate with 0.005 N sodium thiosulfate. A blank titration on 5 ml of the reagent is determined after heating with an equal volume of water.

The difference between the blank and the titration of a determination is equivalent to the copper reduced and thus to the sugar. The corresponding amounts of sugar are given in table 98, p. 603, which is a modified form of the table given in the original article.

Notes. Details of the determination of sugar in 0.2 ml of blood are given in the reference cited.

A 0.005 N thiosulfate solution cannot be kept unchanged for more than a few days. It is advisable to keep a 0.1 N stock solution and prepare 1:20 dilutions as required.

Any agitation of the test tubes, from the beginning of heating in the water bath up to the addition of acid, should be avoided to minimize reoxidation of cuprous oxide by air.

It is undesirable to cool below 30° C. If the sample contains more than 1 mg of sugar, incomplete oxidation of copper may occur.

In order that the pH of the reagent may remain unaltered, it is important that the sample be neutralized with sodium hydroxide (not carbonate). Phenol red is a suitable indicator and renders the end point of the thiosulfate titration more distinct.

(b) BENEDICT MODIFICATION OF FOLIN AND WU METHOD [46, 47]

Reagents. Alkaline copper solution.-Dissolve 200 g of sodium citrate and 60 g of sodium carbonate in about 800 ml of water. Then dissolve 6.5 g of pure copper sulfate crystals in about 100 ml of water and add to the former solution with agitation. Add 9 g of ammonium chloride, dilute to 1 liter, and mix.

Tungstic acid color.-Dissolve 100 g of pure sodium tungstate in about 600 ml of water in a liter flask. Add 50 g of pure arsenic pentoxide, then 25 ml of 85-percent phosphoric acid and 20 ml of concentrated hydrochloric acid. Boil 20 minutes. After cooling this, add 60 ml of commercial formalin, 45 ml of concentrated hydrochloric acid, and 40 g of sodium chloride. Dilute to 1 liter and mix.

To 100 ml of the alkaline copper solution add 2.5 to 3.0 g of pure anhydrous sodium sulfite and preserve for use. This solution is not reliable after 1 month.

Procedure.-Transfer 2 ml of the sugar solution and 2 ml of the copper reagent to a Folin and Wu sugar tube. Into another tube transfer 2 ml of a standard dextrose solution containing 0.1 (or 0.2) mg per milliliter and 2 ml of copper reagent. Mix by side-to-side shaking and place the two tubes in boiling water for 5 mintues. Cool by immersion in cold water and add to each 2 ml of tungstic acid color reagent. After 1 to 2 minutes dilute to 25 ml, mix thoroughly, and compare in a colorimeter. The sugar in the unknown solution is calculated by the formula

Depth of column of standard
Depth of column of unknown'

X10 (or 20)=mg per 100 ml.

(c) FERRICYANIDE MICROMETHOD OF HAGEDORN AND JENSEN

Hagedorn and Jensen [48] devised a method particularly for the determination of blood sugar, but which is capable of extension to the analysis of any material containing minute quantities of reducing sugar. The maximum weight of dextrose which can be determined is 0.385 mg.

When ferricyanide in alkaline solution is heated with a reducing sugar it is reduced to ferrocyanide, the amount of reduction being a measure of the amount of sugar taken. The quantity of reduced ferricyanide is determined as the difference between the total and that remaining after the reduction reaction. The determination depends upon the reaction

2H3Fe(CN)6+2HI=2H,Fe(CN)6+I2,

which is reversible but can be made to run quantitatively from left to right by the precipitation of ferrocyanide as a zinc complex:

2K4Fe(CN)6+3ZnSO4=K2Zn3[Fe(CN)3]2+3K2SO4.

The liberated iodine is then titrated with standard thiosulfate. The Hagedorn-Jensen method has been used extensively in blood analysis. It is introduced into this circular because of its probable general applicability to other materials than blood. It has the great advantage that ferrocyanide is stable in air and no back oxidation occurs. Miller and Van Slyke [49] point out the disadvantage of the determination of ferrocyanide by difference, and they describe a method of direct titration with ceric sulfate, the end point being determined in the presence of the indicator, Setopaline C (a trade name). However, ceric sulfate gave uncertain results with fructose, which curtails the general applicability of the modification. Shaffer and Williams [50] have described a modification in which the amount of reduction is determined by measuring the potential of the ferriferrocyanide electrode. This is extremely rapid and convenient, but is to be recommended only if the number of analyses is great enough to justify the labor of assembling the necessary apparatus. Van Slyke and Hawkins [51], using the Van Slyke-Neil apparatus [52], determined unreduced ferricyanide gasometrically by measuring the pressure of nitrogen evolved by the reaction

4K3Fe(CN)6+N2H1+4NaOH=4K,NaFe(CN)6+4H2O+N2 and have shown that the method can be used rapidly and conveniently for the determination of sugar in both blood and urine. Folin [53] determined the ferrocyanide by a colorimetric measurement of the prussian blue found by addition of ferric chloride to the reaction mixture after completion of the reduction. He also described an effective method for the purification of ferricyanide. Hawkins [54], using the gasometric method, found that the reduction was proportional to concentration for all sugars analyzed except fructose, arabinose, and xylose, in which instances it was proportional up to 0.1-mg concentration. The relative reducing powers are glucose, 1.00; mannose, 1.014; galactose, 0.792; fructose, 0.986; arabinose, 0.949; xylose, 1.019; maltose, 0.725; and lactose, 0.726.

Reagents. (1) Dissolve 1.65 g of K,Fe(CN), recrystallized and dried at 50° C, and 10.6 g of anhydrous Na2CO3. Fill to 1 liter and preserve in the dark.

(2) ZnSO4.7H2O, 10 g; NaCl, 50 g; and water to 200 ml. Immediately before use, add solid KI in the proportion of 2.5 g per 100 ml to the volume of solution required for the analyses. If added to the stock solution, a separation of iodine occurs with lapse of time. (3) Acetic acid, 3 ml of glacial acid in 100 ml.

(4) One gram of soluble starch in 100 ml of saturated NaCl solution. (5) Sodium thiosulfate 0.005 N. Dissolve 0.7 g of crystals and make to 500 ml. Standardize frequently by titration against 0.005 N KIO, (0.3566 g in 2 liters), adding for each 10 ml of iodate solution about 20 mg of KI and a few milliliters of dilute HCI.

Procedure. Into large test tubes (30 by 90 mm) transfer a sugar solution containing less than 0.38 mg of dextrose, and add water to make a volume of 12 ml. Add accurately 2 ml of the alkaline ferricyanide solution and heat in a boiling-water bath for 15 minutes. Cool and add 3 ml of the potassium-iodide-zinc-sulfate solution and 2 ml of acetic acid. Titrate the liberated iodine, using a microburette and 2 drops of starch indicator.

Conduct a blank determination in the absence of sugar. Refer to table 99, p. 604.

Example.-A sample of sugar required 1.26 ml of thiosulfate, and the blank, 1.86 ml; 2.00 ml of standard iodate required 1.90 ml of thiosulfate. Each titer is multiplied by 2.00/1.90 to reduce it to a 0.005 N basis. Corrected titers are, respectively, 1.33 and 1.95 ml. In table 99, 1.33 ml is equivalent to 0.119 mg, and 1.95 ml to 0.008 mg of dextrose. Therefore, the corrected sugar content is 0.111 mg.

6. COLORIMETRIC AND VISUAL METHODS

(a) POT METHOD OF MAIN FOR DETERMINATION OF REDUCING SUGARS IN RAW SUGARS AND SIMILAR PRODUCTS [55]

Main has devised a visual method for the determination of reducing sugars, which is capable of yielding results of high precision and of extending the range of quantitative estimation to very low percentages. The reaction is carried out in large resistant-glass test tubes, 150-mm length by 38-mm internal diameter and weighing 50 to 55 g. In order to avoid back oxidation by air, a series of floats is constructed of similar test tubes having slightly smaller diameter and which make a sliding fit into the others. The barrels of these floats are 100 mm long, and the upper end of each is drawn out to a taper, making a total length of about 170 mm.

The water bath is an ordinary oval iron kitchen pot, tinned inside, the capacity of which is 3 gallons. An overflow is fitted near the upper edge of the boiler, and hot water is added continuously through a "sight feed" to replace loss by evaporation. The temperature of the water must be maintained at the boiling point, for which a large ring gas burner is necessary. While in the water bath, the tubes are supported symmetrically by clips in a carrier.

Two alkaline copper reagents are employed: solution I, for concentrations of invert sugar extending up to 16 percent; solution II, for samples containing a maximum of 0.832 percent.

Solution I.-The usual Soxhlet modification of Fehling solution (p. 170). 10 ml of the mixed reagents are used for the analysis. The results are referred to in table 100, p. 605.

Solution II.-A Soxhlet solution (A) as usual, containing 34.639 g of CuSO4.5H2O in 500 ml of Soxhlet solution; (B) 173 g of Rochelle

salt, 50 g of NaOH, and 14.647 g of K4Fe(CN), in 500 ml of solution. For solution II, mix 1 volume of (A), 1 volume of (B), and 2 volumes of 5 N NaOH. This mixed reagent was designated "L. F. S.", and the results are referred to in table 101.

In the extra-alkaline L.F.S. solution, the ferrocyanide, which is present in the ratio of 1 mole to 4 moles of cupric sulfate, has the function of combining with the reduced copper to form cuprous ferrocyanide and obviates the red coloration that tends to mask the end point as indicated by the methylene blue.

For the standardization of copper solution I, invert sugar is prepared according to the method of Lane and Eynon [36]. Prepare a standard invert-sugar solution by dissolving 9.5 g of pure sucrose, accurately weighed, in about 80 ml of water, adding 5.3 ml of hydrochloric acid (sp gr 1.16, or approximately 10 N), and completing the volume to approximately 100 ml. Allow this solution to stand at 22° to 25° C for 3 or 4 days and dilute to 1 liter. From this stock solution, pipette 50 ml into a 250-ml flask, neutralize with sodium hydroxide (approximately 2.5 ml of N NaOH), and complete to volume. This solution now contains 0.002 g of invert sugar per milliliter.

Transfer the following solutions to each of three tubes in the order stated: 10 ml of mixed Soxhlet solution (solution I); standard neutralized invert-sugar solution, 24.5, 25.0, and 25.5 ml, respectively; and 2 drops of 1-percent methylene blue.

Mix the contents of each tube by gentle rotation and insert the floats so that they rest on the liquid, care being taken not to entrap any air bubbles. Place the tubes in the carrier and immerse in the briskly boiling water for exactly 5 minutes. Then remove and inspect. In general, one of the three solutions will show complete reduction of copper, while the adjacent one will show a trace of blue. The volume of invert intermediate between these two is taken as equivalent to 10 ml of Soxhlet solution. The precision of standardization and also of analysis can be greatly increased by lessening the intervals between the volumes in the tubes. The mean between the last blue and the first red is always taken as the true result, unless the blue color is actually seen to fade in a tube on removing it from the pot at the end of the 5 minutes. In such case, the actual volume in that tube is taken as the correct figure.

As in other methods, some preliminary idea of the amount of invert sugar present in the sample must be ascertained by using a rough incremental titration or other guide.

For the estimation of small percentages of invert sugar, that is, less than 0.8 percent, L. F. S. solution (solution II) may be used. The table upon which the use of this solution is based overlaps table 100 for solution I, since the latter permits the estimation down to about 0.3 percent. The L. F. S. solution should be standardized against an invert-sugar solution containing 0.025 g per 100 ml; 37 ml of such a solution should decolorize 4 ml of L. F. S. in the presence of 2 drops of methylene blue.

The time of heating for amounts of invert sugar below 0.01 percent must be increased to 10 minutes, as shown in table 101, p. 606.