of a candy test, any quantitative statement of the stability of the sample must depend upon analytical procedure and other physical measurements. Examples are the estimation of sucrose hydrolysis (as by polarimetric or copper-reduction methods), the estimation of caramelization (as by measurements of the color in the solid plaque or in a water solution of the candy), and the estimation of the "slump," or specific area, all as outlined above.

Appraisal of the differences in the stability of the various samples, and the separation of the samples accordingly into various classes, is practicable only to the degree that the particular candy-test procedure approximates a reliable reproduction of a constant program of operating conditions each time it is applied. This is to say that the variability among the programs established in numerous individual tests on random samples of the same lot of sugar must result in much smaller differences in decomposition (as distinguished by the chosen methods of examination) than the variations in decomposition arising from the differences in stability of the different lots of sugar that are to be separated on the basis of quality. The tolerances for deviation from a constant program must be smaller the more precise the separations are to be. On the other hand, the minimum tolerances which may be imposed are limited by the character of the candy-test procedure available.

The minimum feasible tolerances which may be stipulated for candy tests according to the Hooker procedure are so great that satisfactory separations of samples on a quality basis are impracticable. This is doubly evident where different operators are involved, what with inadequacy of directions as to arrangement of apparatus and what with the inclusion of details of procedure which inherently are subject to individual variability. No long array of samples could be arranged precisely in the order of minutely graduated differences of stability, nor could the degree of departure of a sample from an established standard of quality be indicated, on the findings of candy tests carried out by different operators by means of the Hooker procedure alone or by means of any of its ordinary modifications. The deviations are so great, indeed, that nearly every user of the test has entertained at least transient doubts that the results obtained even by a single operator are valid criteria for any useful separation of samples on a stability basis [11, 12]. The circumstances are complicated by the failure of many users to appreciate the fact that no finer separations should be demanded than definitely will be put to practical use.

Very pure sucrose in candy tests is distinctly weak but relatively stable in respect to caramelization. The purest commercial products have similar characteristics in both respects. Because of this, certain brands of "first-run" char-refined sugar, as produced for confectioners' use, have been strengthened at the refinery for many years by the addition of small quantities of sodium carbonate or sodium bicarbonate 33 to the water used for washing the crystals [5]. The small amount of impurity thus deliberately introduced has but little influence upon the color of hard candy produced from the sugar, although much excess of the alkali would bring about marked caramelization.

23 In 1897 Wiechmann presented data [1] on experimental suppression of sucrose hydrolysis by the presence of 1 part of calcium oxide, sodium carbonate, or sodium bicarbonate, in 100,000 parts of refined cane sugar during its conversion to barley sugar. Possibly even earlier, Hooker initiated the practice of strengthening confectioners' sugars, in response to the complaints of certain consumers. The treatment was devised and originally controlled through the Hooker candy-test procedure.

The stability of refined cane sugar of ordinary quality is inferior to that of the strengthened confectioners' sugars in respect to both hydrolysis and caramelization [5]. On the other hand, beet sugars generally tend to be distinctly stronger than either type of cane sugar but definitely less stable than either as to caramelization. There are notable exceptions to these usual relations of the three types of sugar mentioned. Indeed, the stability gradations of all sorts are so nearly continuous that probably no one would propose to distinguish the types of sugar to which many individual samples belong, solely on the evidence of candy tests.

Because such gradations are continually encountered, most users and prospective users of candy tests need a method which is both more reliable and more precise than any of those heretofore available. For selecting or for segregating different lots of sugar according to the requirements of different consumers, without an excessive number of tests, a high degree of reliability is demanded of the individual candy test. For the precise control of quality in sugar production in any plant with relatively little variation of stability in its products, the candy test procedure should have a high degree of reproducibility_to permit stipulation of small tolerances of standard deviation. For many other purposes, like precision must be attained. The method and equipment described for the National Bureau of Standards simple barley-sugar test procedure are designed to reduce the standard deviations, as compared with those which prevail in the candy tests heretofore available, (1) by diminishing the variations of practice through more specific directions, (2) by diminishing the influence of the operator's individual technique upon the procedure, and (3) by greatly reducing instrumental variability in several important respects. Providing these specifications are adhered to in detail both as to operation and equipment, this enhancement of precision in candy tests should be realized in fact. With attention to the order of occurrence of stability variations in relation to the usual control and operating data of a plant, as a clue to the presence or to the advent of "assignable causes", manufacturers can make use of the increased precision as a means of bringing the production of sugar more definitely within the field of statistical control as practiced in certain other industries. The dependability and increased convenience of the new procedure, once the equipment is installed, and as compared with what heretofore has been available, should place the candy test on a much surer footing from the viewpoint of the consumer as well as tho producer.

3. CONCLUSION

After quoting Hooker's-directions for candy-test procedure for comparison, directions for a new and greatly improved simple barleysugar test method are presented in detail, together with specifications for the apparatus required for its use. It is proposed not only as a standard method for the inspection of commercial sucrose with respect to stability to heat, but also as a model procedure which with suitable modification can be converted to various other types of candy tests for the examination of all sorts of sugar products. Details of the latter use are reserved for discussion in another place. It is pointed out that the method as used for the examination of commercial sugar is of value both to the producer who wishes to establish a more effective

control of the quality of his product, and to the consumer who either purchases sugar on quality specifications or allocates his purchases to different uses according to their quality. The possibility has been suggested of implementing the control of quality in the production of sugar through the introduction of modern statistical methods which make use of the order of occurrence of stability variations as a clue to their assignable causes.

4. REFERENCES

[1] F. G. Wiechmann, J. Phys. Chem. 1, 69–74 (1897).

[2] S. J. Osborn, The Use of Beet Sugar in Candy-Making (1913), unpublished report of the techincal staff of the Great Western Sugar Co., Denver, Colo. [3] Cir. BS C44 [ed. 2] (1918).

[4] Max J. Proffitt, Laboratory Work in Connection with Experiments on the Production of Candy Sugar (1914); Filtration Experiments (1916); unpublished reports of the technical staff of the Great Western Sugar Co., Denver, Colo.

[5] Frederick W. Murphy, Candy Mfr. 1, [3] 36–37, 51 (1921).

[6] H. S. Paine, M. S. Badollet, and J. C. Keane, Ind. Eng. Chem. 16, 1252–60 (1924).

[7] J. A. Ambler, Mfg. Confectioner 7, [1] 17-19, 82 (1927).

[8] G. L. Spencer and G. P. Meade, Handbook for Cane Sugar Manufacturers and Their Chemists, p. 189–90 (John Wiley & Sons, New York, N. Y., 1929).

[9] J. A. Ambler and S. Byall, Ind. Eng. Chem., Anal. Ed. 7, 168-73 (1935). [10] Otto Windt, Mfg. Confectioner 12, [8] 20-24 (1932).

[11] A. B. Kennedy, Mfg. Confectioner 13 [5] 18–23, 51 (1933).

[12] E. K. Ventre and S. Byall, Report of Studies on Uniformity of Quality of Sugars, U. S. Dept. Agr. Bur. Chem. Soils (1938).

[13] Max J. Proffitt, Unpublished paper on candy test methods.

XXV. PREPARATION AND PURIFICATION OF PURE

SUGARS

1. DEXTROSE

With a view to assisting in the unification of methods of sugar analysis, the National Bureau of Standards issues chemically pure dextrose as one of its standard analyzed samples. This was first issued in 1914. At that time, and for some years thereafter, the method of purification and crystallization was that devised by Jackson [1]. The raw material used was the purest form of commercial dextrose obtainable at that time and was marketed under the trade name of Cerelose. It was a brownish-colored granular material containing about 87.2 percent of dextrose, 3.9 percent of nonfermentable sugars, 0.55 percent of ash, and a quantity of dextrins. The method is as follows:

34

The impure material is submitted to a preliminary treatment similar to that described by Bauer [2]. The crude material is digested in a shaking machine with alcohol in order to wash the mother liquor from the crystals. The mass is then poured into a centrifugal machine and drained at high velocity. The drained crystals are then washed repeatedly in the rotating basket with fresh portions of alcohol. The washed substance is dissolved by heating in 40 percent of its weight of water and 140 ml of alcohol added for each 100 g of the washed substance. This mixture is then heated on a steam bath and filtered to remove the precipitated impurities. The filtrate is evaporated in

At the present time this name is applied to a highly purified commercial dextrose having the physical appearance of granulated beet or cane sugar.

the vacuum boiling apparatus described on page 393 to about 60 percent of sugar. The supersaturated liquid is then transferred to a shaking machine, a few crystals of dextrose hydrate added, and the substance allowed to crystallize in motion. After standing overnight, the crystal mass is centrifuged and washed four to five times with redistilled alcohol. The crystals obtained are very white and already possess a high purity. They are then dissolved in water, the solution filtered through asbestos into the boiling apparatus, concentrated, and recrystallized from the aqueous solution. The second crystals are carefully dried and analyzed for moisture, ash, polarization, and reducing power. If these tests indicate the presence of any remaining impurity, the recrystallization is repeated until a satisfactory product is obtained.

The crystals thus obtained are dextrose hydrate. Formerly in the preparation of standard samples the hydrate was freed of its water of hydration by heating in an electric oven at 60° to 70° C.

The above method of purification was followed for some years, until there became available white crystalline dextrose in commercial quantities. The following modification of the method of purification was then adopted. A quantity of the white crystalline dextrose is dissolved in hot water to make a 60 percent solution. This is treated with char and filtered. The filtrate is concentrated in a glass boiling apparatus under reduced pressure to a sirup of about 80 percent of dextrose. The thick sirup, which is brought to a temperature of 55° C, is seeded with crystals of anhydrous dextrose and placed in a rotating crystallizer enclosed in an insulated, electrically heated cabinet held at 55° C. The crystallization is usually complete within a few hours. The temperature is reduced until the massecuite has reached the temperature of approximately 30° C, when the crystals are separated on a centrifugal machine and washed with alcohol. The crystallization is repeated until analysis shows the material to be of the requisite purity. The air-dried crystals of anhydrous dextrose, thus obtained, contain less than 0.1 percent of moisture. The material is finally dried in vacuum desiccators containing suitable drying agents, with or without the aid of heat, to a moisture content of 0.01 percent or less. This purified material is issued as Standard Sample 41, Dextrose.

In the preparation of pure dextrose hydrate the 60-percent solution, which has been treated with decolorizing carbon and filtered as described above, is seeded with crystals of dextrose hydrate and rotated in the crystallizer at room temperature until crystallization is complete. The crystals are centrifuged, washed with alcohol, and dried in the usual manner.

The standard dextrose sample may be used as a standard in methods for the determination of reducing sugars to replace the usual invertsugar solution and has the advantages of greater convenience in the preparation of the solution and greater certainty of composition. Some uncertainty arises in the preparation of invert-sugar solutions on account of the method of inversion employed. Reversion products may be formed, or either levulose or dextrose injured by too prolonged action of the acid. Furthermore, if the solution is preserved, decomposition may occur if either acid or alkali is in excess. The standard dextrose solution may be prepared very readily and conveniently as

needed by weighing out the required amount of solid substance and dissolving in the necessary volume of water.

2. SUCROSE

Most of the methods for the purification of sucrose which have been described depend upon the use of alcohol as a precipitant. Sucrose is almost insoluble in absolute alcohol, and its solubility in alcohol-water mixtures diminishes very rapidly as the concentration of alcohol increases.

(a) METHOD OF HERZFELD [3]

Prepare a cold, saturated, filtered, refined sugar solution and with continual stirring add an equal volume of 96-percent alcohol. Filter after 15 minutes, wash with ether, and dry in a water bath.

(b) METHOD OF HERLES

This method gives particular attention to the elimination of raffinose. To exclude traces of raffinose, precipitate a cold, saturated, filtered solution of refined sugar with alcohol, warm on a water bath to 60° C, decant the solution, pour fresh alcohol on the precipitate, warm while stirring, decant again, wash the sugar on a filter with absolute alcohol, and dry in a thin layer on filter paper at 30° to 40° C. On account of the well-known solubility of raffinose in methyl alcohol, many operators have used this as a precipitant to insure the removal of this impurity.

(c) METHOD OF THE INTERNATIONAL COMMISSION FOR UNIFORM METHODS OF SUGAR ANALYSIS [4]

To prepare pure sugar, further purify the purest commercial sugar in the following manner: Prepare a hot, saturated, aqueous solution, precipitate the sugar with absolute ethyl alcohol, spin the sugar carefully in a small centrifugal machine, and wash in the latter with absolute alcohol. Redissolve the sugar thus obtained in water, again precipitate the saturated solution with alcohol, and wash as above. Dry the second crop of crystals between blotting paper and preserve in glass vessels for use. Determine the moisture still contained in the sugar and take this into account when weighing the sugar which is to be used.

(d) METHOD OF BATES AND JACKSON [5]

A method of purification has been developed at this Bureau in which recrystallization from aqueous solution is utilized. The purest granulated sugar of commerce is dissolved in an equal weight of water. This relatively dilute solution is clarified from albumenoids and suspended material by the addition of alumina cream which has been carefully freed from soluble salts by continued washing and testing the wash water with barium chloride. The solution of sugar is filtered by pouring on large, fluted filters of hardened filter paper. The filtered solution, which usually has the brilliancy of distilled water, is boiled in a vacuum-distilling apparatus at a temperature not exceeding 35° C until the concentration of the solution, which was initially 50 percent, reaches 76 to 80 percent.

The requisites for a serviceable concentrator may be briefly summarized as follows: (1) Cleanliness, (2) low boiling point, (3) large capacity, and (4) filtration of entering sirup.