preferable to adopting a different standardization temperature, such as 25° or 27.5° C, as sometimes suggested.

Einsporn and Schönrock [4] have made an elaborate study to find by calculation what the corrections would be if the polarization is carried out at the tropical temperatures of 25° and 30° C. Their final values are identical with those which this Bureau has always given for these temperatures in its tables of temperature corrections accompanying quartz control-plate certificates. (See fig. 23.)

If the solution is made up to volume at 20° C, but is polarized at another temperature, all apparatus being at that temperature, equation 32 is no longer applicable. Under these conditions the temperature coefficient of the normal sucrose solution alone is given by Schönrock as 0.000469, while that of the saccharimeter, as previously shown, is 0.000148. If a glass tube is used, we have as the total temperature correction

0.000469-0.000008+0.000148 0.000609.

The polarization in sugar degrees at 20° C (S20) of the near normal solution of sucrose is therefore given by

S20

St+S20.000609 (t−20° C).

(34)

Equation 34 obviously holds only while the number of grams of sugar in 100 ml of the solution remains unchanged.

(d) SUGAR MIXTURES

The coefficient 0.0003, in eq. 32, having been determined for the normal weight of sucrose, should be applied with considerable care. It has, however, long been a widespread practice to apply it to all sorts of saccharine products, with the result that a polarization, supposedly accurate, may contain errors of appreciable magnitude. If the solution does not read approximately 100° S, the correction to give the reading at 20° C should not be obtained by multiplying 0.03 by the difference in temperature-a common practice. In general, no great error will be introduced by following this procedure provided the polarization is above 85° S. Nevertheless, the better and safer practice is to solve eq 32, thereby correcting for the difference in sucrose concentration from the normal solution.

An even more widespread practice has been to apply eq 32 to sugar polarizations without regard to the associated impurities. This is particularly true of raw sugars and molasses which contain appreciable quantities of invert sugar. Of the constituents of invert sugar, the dextrose has a negligible temperature coefficient, while the levulose has a very large coefficient, and in such a direction that the positive rotation of the mixture tends to increase upon elevation of temperature. It is therefore manifestly erroneous to apply a pure sucrose temperature coefficient to a mixture unless all the substances, except sucrose, are present only in such small quantities that the error introduced is negligible.

For the temperature correction of the better grades of raw sugar, the temperature formulas 32 and 34 give results which are sufficiently accurate; but if they are applied to low-grade products, errors are

introduced. Raw sugars may be considered as mixtures of pure sucrose and cane molasses. To correct the whole mixture for the effect of temperature change, it would be necessary to apply the resultant coefficient obtained by combining the separate coefficients for sucrose and for molasses, taking into consideration the relative quantities of the two and the constitution of the particular molasses which contaminates the sample. C. A. Browne [5] has done this and shows that the temperature coefficient varies in almost direct proportion to the content of molasses. It is in general incorrect to apply this computed correction because of possible individual variation in the constitution of the molasses. Accordingly Browne advises the omission of the temperature correction for low-grade samples. He has computed an average coefficient for large numbers of samples of raw sugar which may be determined according to the polarization. Thus for cane products the coefficient is 0.0015. It is seen from this that at 80° the coefficient becomes zero and for lower grades it becomes positive instead of negative. If individual variations did not occur, this correction would be a useful one; but as it stands, it merely serves to show the possible error of applying eq 32 to low-grade sugar. For beet products the coefficient is 0.0006.

4. THERMOSTATS
(a) WATER

The satisfactory control of temperature in polarimetric work is an important and troublesome subject. For accurately making up solutions to volume at a desired temperature, water thermostats which give extremely close regulation are to be preferred. Many types which are entirely satisfactory are available from apparatus dealers. These consist essentially of metal or glass tanks suitably insulated on the outside and nearly filled with water. The cooling is accomplished by circulating ice water through an immersed coil, and the temperature regulation is maintained by immersed lamps or heaters operated intermittently by means of mercury relays connected to mercury thermoregulators. The water is kept in constant motion, insuring an even temperature throughout, by means of electric turbines or stirrers. For circulating constant-temperature water through the jackets of polariscope tubes, refractometers, etc., use is made of small electric-driven pumps.

(b) AIR

At the National Bureau of Standards most polariscopic measurements are made in a constant-temperature room maintained at the standard temperature of 20° C. Three such rooms are available. The largest of the three, used for both routine and research work, is approximately 10 by 20 feet. The walls and ceiling are insulated with thick corkboard. The temperature is maintained by the automatic. intermittent operation of a large ammonia compressor located in the attic room above, the direct-expansion ammonia coils being mounted on the side walls of the constant-temperature room. Satisfactory temperature control is accomplished through the use of a bimetallic regulator.

The saccharimeters are placed within the room on a table of convenient height. For special work the instruments are enclosed in

individual insulated boxes. This so retards any change in temperature. of the instrument that, when the room is operating within a few tenths of a degree, no significant change can be detected on the thermometer inside the instrument case after equilibrium is reached. The illumination of the instrument is secured by placing suitable light sources just outside the room, allowing the light to pass into the room through small glass windows in the walls.

One of the rooms used for special research work, in which the Bureau's large polariscope, described on page 46, is located, is approximately 7 by 10 by 9 ft. with an adjoining closet in which is housed refrigeration coils feeding from the central refrigerating system

RO

F

7'

7'

(CO2) of the Bureau and controlled by a needle valve on the intake side (fig. 24). A fan, F, located in an opening near the top of the closet blows cold air from around the refrigeration coils, RC, out into the room into the space above a false ceiling, C, which 10' is perforated throughout and serves to distribute the cold air evenly over the top of the room. An opening near the bottom of the closet allows return air to be sucked into the closet to be recooled. Cold air is prevented from circulating in reverse direction, when the fan is not operating, by means of a valve, V, consisting of a flap of heavy cloth tacked across the upper edge of the bottom opening. When the fan stops, the cloth falls against the opening and effectively seals it against the cold air in the closet.

9'

FIGURE 24.--Constant-temperature

room.

Thus cold air is fed into the room only while the fan is operating and in an amount depending upon the speed of the fan.

A heater, II, attached to the lower side of the false ceiling may be operated when needed, either continuously by a switch or automatically by a thermostat. This heater is used, however, only when temperatures above outside room temperature are required.

Upper, basal section; lower, vertical section.

The fan which controls the amount of cold air put into the room is operated automatically by a bimetallic-strip regulator through an electronic relay. This avoids the use of all mechanical relays and their troublesomeness.

The assembly used for operating the fan is shown in figure 25 and is largely self-explanatory.

It is essentially a full-wave grid-controlled rectifier arranged for on and off or relay operation. It was assembled from stock parts which were immediately available, with a view to extreme flexibility and adaptability to other uses. Had it been engineered only for the par

ticular job in hand, considerable simplification could have been made. For instance, the variac, T1, could have been dispensed with and also a much smaller transformer of the correct voltage and power rating used in place of T2. However, the apparatus as assembled has proved very convenient and dependable in operation and has a power-handling ability of over 500 watts.

In operation, current is first switched onto the filament lighting transformer, T., and the filaments of the FG 57's allowed a heating time of about 5 minutes before switching on the plate power, T2. For the sake of simplicity of the drawing, the filaments are shown lighted by separate transformers. In practice they may just as well be paralleled on a single 5-volt winding.

After the filaments have reached their operating temperature, the switch, S, is closed, activating the variac, T1, whose output tap

FIGURE 25.-Wiring diagram for electronic-relay control of thermostated cold room. TI, GR Variac, 200 CM; output voltage, 0 to 130 volts; T2, GE transformer No. 9TM416A; primary 683 kilovolt amperes, secondary 990 kilovolt amperes, 110/220/55 to 550/275 volts; Ts. Filament transformer, 110 to 5 volts, 50 watts; S, Fuse and switch; VT, Thyratrons FG 57; plate current, 2.5 amperes each, maximum average; 1,000 volts, maximum peak; lament, 5 volts, 4.5 amperes each; R1, 20-ohm resistors (660watt heater units with medium screw base) mounted in porcelain lamp-socket bases; R2, 50,000-ohm, 1-watt resistors; R3, 100,000-ohm, 1-watt resistor; R., 10,000-ohm, 1-watt resistor; C, 1-microfarad capacitor; B. 7.5-volt bias battery; K. American Instrument Co. "Quickset" bimetallic regulator; V, Directcurrent voltmeter, for use when adjusting the load voltage by means of T1; O, Hubbell plug which serves as output terminals; F, Fan located in cold closet; and W, Twisted lamp cord leads.

has been set at or below the proper operating position for the particular load, which is connected through a Hubbell plug at O. When first adjusting the apparatus, the voltage on the primary of T2 is varied by means of the variac, T1, until the direct-current voltmeter, V, shows the correct voltage across the load at 0, which in this case is the fan motor, F. Thereafter the variac may be left at its proper setting and the power simply switched on at S.

For the particular fan at present in use, about 70 volts is required on the input side of T2 to maintain the fan's rated voltage of 120 volts at O.

The resistors, R1, are ballast resistors to protect against current surges and extreme overload on the tubes made possible by the nonresistive character of the load (fan motor). Even if the load were short-circuited, R, would limit the current to the point where the tubes and transformer would not be injured before the fuse could blow or an overload relay, if used, could operate. Heater units, 110-volt 660watt, such as are used in the ordinary household electric reflecting heater, serve admirably. They consist of resistance wire wound on a ceramic cone which carries a screw plug on one end similar to that on ordinary incandescent lamps and which fits the ordinary medium screw-base lamp socket.

Load current is not passed by the tubes as long as the grids are kept sufficiently negative. When the contacts of the bimetallic regulator, K, are open the grids assume practically ground or cathode potential and load current normally flows, operating the fan. When sufficient cold air has been blown into the room to cause the contact, K, to close, the battery, B, operating through the networks R2RR, impresses a negative bias on the grids sufficient to block off the plate current and stop the fan. Since this battery, which is an ordinary 7%-volt radio C battery, is only connected in the circuit intermittently,

and then through a high resistance of more than 100,000 ohms, its expected life is not much less than its shelf life.

The condenser, C, serves to stabilize the grid-cathode difference of potential and at the same time, in connection with R, and R3, serves as a time-delay device, to delay starting or stopping until after K makes good contact or breaks contact. This aids in eliminating chattering due to mechanical vibrations when the contacts, K, are nearly closed. R3 serves to discharge C after K opens, thus allowing the grid to return to cathode potential and start the fan. R, serves to limit the rush of charging current to C when K closes, preventing sparking at the contact, K. By use of this device the temperature of the room may be maintained constant at any desired temperature within the range of the cooling system, for long periods of time, with a variation of only a few tenths of a degree. The exterior walls of the room are corkinsulated. Provision also is made for cooling the room in wintertime by the use of outside air instead of refrigeration, by blowing the cold