stainless-steel weights have been introduced, and the indications are that they will prove satisfactory if made from steel of the proper composition. The working standards should conform to the Bureau's requirements for class B and the reference standards (i. e., those used only for checking the working standards) should conform to the requirements of class A. The maximum error allowed in both classes is 2 mg on the 26-g weight and 1.5 mg on the 13-g weight.

(b) ANALYTICAL

In most analytical procedures in the sugar laboratory, a good grade of analytical weights should be accurate enough, but it is not safe to rely on them unless they are tested. For work requiring considerable accuracy in the weighing, the weights should conform to the Bureau's requirements for class S. The maximum errors allowable range from 0.5 mg on the 100-g weight to 0.1 mg on the 1-g weight and 0.02 mg on the 10-mg weight. For less accurate work, weights conforming to the requirements of class S2 having allowable errors of five times those of class S, may be used.

(c) REFERENCE STANDARDS

Weights are liable to change. They cannot be used without a certain amount of wear, which will ultimately make an appreciable change in their values. Ordinary analytical weights sometimes suffer serious change from the oxidation of adjusting material placed in the cavity under the knob. Weights must therefore be retested from time to time according to the nature of the weights and of the work for which they are used. Reference standards are therefore needed, since it is seldom advisable to send the weights to this Bureau as often as would be needed. There is no gain in the purchase of complete sets for this purpose when this is done at the expense of quality, as must usually be the case. A set of working standards can be tested readily by intercomparison of the weights among themselves, if one or two reference standards are available on which to base the calibration. The Bureau will furnish information in regard to series that are much better than the ones generally used and yet involve no great amount of additional labor. Probably the best denominations for reference standards would be one 100- or 50-g weight, one 1-g weight, one 10-g weight, one 26-g weight, and one 13-g weight. The 1-g weight should be made of platinum, as it is the starting point. for the determination of the milligram weights.

For the best reference standards, the Bureau recommends one-piece weights. Gold- or platinum-plated Tobin bronze weights are the most satisfactory ones now available for this purpose for weights above 1 g. Weights having a bard metal driven plug would rank Standards for the analytical sets should conform to the requirements for class S and be tested under that class.5 Reference standards for the "sugar weights" should come under class A, as stated above.

(d) CERTIFICATION

Sugar weights and analytical weights are among the weights tested by the Bureau, but the rougher weights are not regularly accepted. for test.

The same kind of weights, but less accurately adjusted, may be obtained as class A and class B weights. For extreme accuracy (seldom, if ever, needed in weighings for polarimetry), when careful corrections must be made for the buoyant force of the air, only one-piece weights can be relied upon, and the volume of each weight must be determined. This requires that the weights come under class M.

Information as to the precision to which corrections will be certified for weights of classes A, M, or S, and lists of tolerances for any class, will be supplied on application to this Bureau. Full details as to specifications, tolerances, and precision of corrections, together with other information as to standard weights and some of the methods of testing them, are given in National Bureau of Standards Circular C3.

To assist in the identification of the weights, the test number assigned to the weights by the National Bureau of Standards will be stamped on the bottom of the box provided for keeping them, thus NBS Test No. 4978.

The shipping case or the inner wrappings will always be sealed when tested weights are shipped from the Bureau.

(e) SUBMISSION OF WEIGHTS FOR TEST

A written request for the testing should be sent when the weights are shipped. This should always indicate the class of weights submitted, and if two different tests are available in that class, the character of test desired. Sufficient information should also be given to enable the Bureau to identify the package.

If weights have already been used as standards in exacting work, and it is important to know what their corrections were at the close of such work, this fact should be stated; otherwise, weights are carefully cleaned before being tested.

Weights should be packed tightly. Sets in covered cases generally need extra packing inside the case. The very small sheet-metal fractional weights are especially likely to work out of place and be damaged.

Address packages and correspondence, "National Bureau of Standards, Washington, D. C."

6. BALANCES

The methods of analysis employed in the sugar laboratory require the use of a high-grade analytical balance for such operations as the determination of ash, moisture, and specific rotation, as well as a special sugar balance for the rapid weighing of sugars, molasses, and sirup for the usual saccharimetric determinations.

In selecting an analytical balance, it is well to keep in mind several essential points. The sensibility of the balance is influenced by (1) distance between the center of gravity and the point of support, (2) coincidence of the planes of the three surfaces on which the three knife-edges bear, (3) length of the arms of the beam, and (4) reduction of friction to a minimum by finely ground and polished knife-edges and planes. In addition, the sensibility is affected by the weight of the beam; in general, a balance with a light beam is more sensitive than one with a heavy beam.

The Bates sugar balance, especially designed at the National Bureau of Standards for saccharimetric work, has a number of improved features. The customary bows supporting the pans have been replaced by single-arm hangers at the back, an arrangement which gives free access to the pans and reduces to a minimum the danger of spilling sugar on the pans. The weighing scoop is adjusted to balance exactly either of the pans, thereby avoiding the use of a counterpoise weight.

The errors introduced by evaporation or absorption of moisture during weighing are greatly reduced by the rapidity of weighing with this balance. The general design and construction of the balance is such as to make it especially adapted for use in the tropics. It has a capacity of 200 g, with a sensitivity of 1 mg, which may be increased to 0.2 mg by raising the center of gravity weight on the pointer.

[1] NBS Circular C8, Testing of Thermometers (current edition).

[2] G. K. Burgess, BS J. Research 1, 635 (1928) RP22.

[3] E. F. Mueller and R. M. Wilhelm, Proc. Am. Soc. Testing Materials 38,

pt. 1 (1938).

PART 2. RAW AND REFINED SUGARS AND SUGAR

PRODUCTS

VII. POLARIZATION OF RAW AND REFINED SUGARS

1. SAMPLING AND MIXING OF SAMPLES

The importance of a correct method of sampling raw sugar cannot be overestimated. If this step is carelessly or erroneously carried out, the most careful work on the part of the chemist is vitiated. The details of the process of sampling should be arranged with two requirements kept in view: (1) The sample should be thoroughly representative of the package, and (2) after the sample is taken, it should undergo no change until used for analysis.

In order to obtain a representative sample, the sampling instrument, called the trier, should be plunged into the middle of the package and drawn out filled with sugar. If the trier has the correct length and the sampling is skillfully done, all the layers of sugar in the package are represented in the sample. Great care should be exercised to avoid taking a surface sample, since the variations caused by drying or absorption of moisture have their maximum effect at the surface. The dimensions of sugar triers are given in the regulations of the United States Treasury Department relative to sampling of imported sugars. These regulations are printed beginning on page 781 of this Circular.

If the sugar is contained in barrels or other wooden packages, it should be sampled by running the "long trier" or "barrel trier" diagonally through the package from chime to chime. In case molasses has drained to the bottom of the package, especial care must be taken to obtain the sample symmetrically.

In procuring samples from large shipments of sugar, it is advisable to take samples from every package, to mix the large quantity of material thoroughly, and to resample the mixture for analysis.

In order to prevent a change in the composition of the sample taken, the total contents of each trier should be emptied into a tightly covered receptacle and the trier left clean for the next sample, the whole operation being completed within a few seconds. The subsequent mixing and resampling should be conducted in such manner as to avoid unnecessary exposure to the atmosphere. These precautions are necessary because of the great tendency of raw sugar to change its moisture content when exposed to the air.

A large proportion of the impurities in raw sugar is in the form of molasses clinging to the surfaces of the crystals, and the evaporation or absorption of moisture may be very rapid. The action of an absorbing material during mixing (such as the brown paper frequently used) may affect the polarization very markedly by wiping the molasses from the surfaces of the crystals. If samples are to be pre