much time to the subject. So far as possible, it is desirable that the lines utilized be uniformly distributed throughout the visible spectrum. Lowry [35] has suggested the following (given in angstroms): Lithium, 6708, red; cadmium, 6438, red; sodium, 5893, yellow; mercury, 5461, green; cadmium, 5086, green; cadmium, 4800, blue; mercury, 4539, violet. 6708 and 5893 were obtained from flame spectra, and 5461 and 4359 from the quartz-mercury lamp. The cadmium lines 6438, 5086, and 4800 Lowry suggests be obtained from a rotating arc. The electrodes must rotate in opposite directions at a speed sufficiently high to prevent flickering. As electrodes, he uses an alloy of 28 percent of cadmium and 72 percent of silver. The melting point is 860° C. This method has been used at this Bureau and is capable of giving excellent results. The rotating arc is, however, rather difficult to manipulate. Some of the silver lines, namely 5469,

5209, and 4208, may be obtained in this manner also, and with sufficient intensity for polarimetric measurements.

(2) BATES CADMIUM-GALLIUM LAMP [36].-In order to improve the cadmium source, this Bureau has developed a vacuum type of cadmium arc. If pure cadmium is used in a quartz lamp, the adhesion between the cadmium and the quartz results in the destruction of the lamp upon the solidification and cooling of the cadmium. If the cadmium is mixed with mercury to make a soft alloy, the cracking of the lamp is effectively prevented, but the mercury vapor then carries most of the current, since the vapor pressure of mercury is much greater than that of cadmium.

On the other hand, gallium although melting at about 30° C, boils at a very high temperature, approximately 1,500° C. Its vapor pressure is therefore negligible in comparison to that of cadmium. Furthermore, the addition of a few drops of gallium to 10 or 15 ml of cadmium is found to change completely the texture of the latter, rendering it relatively soft and greatly reducing its tensile strength. Subsequently, it was discovered that upon. distilling the cadmium from the alloy at a pressure of about 0.001 mm of mercury, the minute quantity of gallium carried over was sufficient to change completely the character of the cadmium and to prevent adhesion between the cadmium and the walls of the lamp. The type of lamp usually used is shown in figure 14.

3000

The total volume of the lamp was approximately 10 ml. The electrodes consisted of tungsten wires, B, entering through quartz capillaries. These were closed with seals similar to the type described

FIGURE 14.-A, Bates type of cadmium-gallium arc; B, gallium spectrum. A, Cadmium-gallium alloy; B. tungsten electrode; C, lead seal; D, tungsten lead wire: E, capillary for sealing off lamp from pump.

by Sand [37], and later by quartz-Pyrex-tungsten seals when these became available.

In filling the lamp, cadmium containing 2 or 3 percent of gallium, is placed in the side tube, F, and distilled under a vacuum of 0.001 mm Hg, or better. When carefully prepared, the lamp will have an indefinite life. One of this type has been in intermittent use for several years and shows no sign of deterioration. The lamp may be

started by heating with a flame to vaporize the metal. It will operate with as little as 3 amperes and a corresponding drop of 14 volts across the lamp, but most satisfactory results were secured with a current of about 7 amperes and a drop of about 25 volts across the lamp. It may be connected directly to a 110-volt direct-current power line through a suitable resister. Under these conditions a practically pure cadmium spectrum of great brilliancy is obtained. There are no gallium lines to interfere between 4200 and 6400 A, and the ones that do occur are so faint as to be wholly negligible in polarimetric work.

(e) LITHIUM FLAME

The lithium red line (λ=6708A) may be obtained in satisfactory intensity by blowing lithium carbonate dust into an oxyhydrogen

FIGURE 15.-National Bureau of Standards apparatus for obtaining the lithium

flame.

flame. Apparatus devised for this purpose is shown in figure 15. It consists essentially of a cylindrical glass container, at the bottom of which is a small fan driven by a motor. Lithium carbonate, together with a small quantity of Ottawa sand, is placed in the apparatus. The sand, which is picked up and kept in violent motion by the fan, keeps the carbonate from packing and acts as a sand blast, grinding the carbonate finer and finer the longer it is operated. A slow stream of dried air enters at the bottom and leaves at the top, laden with lithium carbonate dust, whence it is conducted by a rubber tube to the housing around the oxyhydrogen burner.

(f) REMARKS ON PURITY OF LIGHT

In most instances absorption filters do not accomplish sufficient purification. For work even approaching precision, spectral purification needs to be used even with those sources which produce line spectra. The purity of the light used for making rotation measurements has not in the past had the attention which is due it.

If one measures a normal quartz control plate, for instance, with first one and then the other of the two D lines (A=5896 and 5890 A), which constitute the sodium doublet, a difference of about 0.08° is obtained. Using a good polariscope rotation measurements can be made with a precision of about 0.003 circular degree; while even the smaller instruments yield a precision of about 0.01°. It is obvious, therefore, that a monochromaticity approaching 1 angstrom unit is required even for ordinary work, and a considerably greater degree of purity for precision work, if the uncertainty in the rotation because of wave-length errors is to be reduced to the same order of magnitude as the experimental error involved in making the settings on the scale (matching the field) of the polariscope.

3. QUARTZ CONTROL PLATES

Quartz control plates are plates of crystalline quartz designed to be used as standards of rotation to facilitate precise saccharimetric and polarimetric measurements. They are indispensable in standardizing saccharimeter scales, and also in controlling saccharimetric and polarimetric measurements in the field, by checking the over-all accuracy of saccharimeter or polarimeter at the time the measurements are being made.

Inasmuch as the highest possible precision is frequently called for in polarimetric measurements, quartz control plates must be designed, constructed, and standardized in a manner commensurate with necessary accuracy and dependability.

(a) REQUIREMENTS AND METHODS OF TESTING

(1) CRYSTALLINE PURITY.-First among the requirements is that of purity; the plate must be of optically homogeneous quartz and contain no striae, inclusions, twinning, or other flaws, which might render the plate unreliable in service. Such flaws, even if not within the effective aperture, i. e., flaws which occur only around the edges covered by the mounting, are not permissible, since the changes in temperature may cause differential expansion and set up strains in the plate, which might make it of doubtful utility.

Plates are tested for purity by placing them between large accurately crossed nicols in a darkened room, using an intense white-light source and compensating for the rotation of the plate by means of a quartz compensating-wedge system. Flaws may best be detected by focusing the observing telescope sharply upon the plate and then rotating the plate in its own plane. Any flaw in the plate will be seen to move with the plate, and hence will be readily detected. With proper technique, this can be made an exceedingly delicate test.

(2) PLANENESS AND PARALLELISM OF THE FACES.-A second requirement is that the faces of the plate shall be both plane and parallel to a sufficient degree of precision; otherwise the plate would