DESIGNS OF MODERN ULTRACENTRIFUGES

E. B. Kofman, Moscow

Repeated attempts to observe sedimentation in highly dispersed sols by means of a laboratory centrifuge have not yielded positive results. The principal cause of these failures is the convection currents that continually arise in centrifuges as a result of the nonuniform heating of the rotating liquid through friction of the receivers against the air and friction in the bearings. Because of mixing, the rate of settling proves to be considerably less than expected, and the disturbance of the proper sedimentation of small particles makes quantitative measurements impossible. As the centrifugal force increases—even with unchanged temperature gradients—the convection currents should intensify; in actual fact, the nonuniformity of heating is also increased. Therefore increasing the rotational speed of a centrifuge above a certain limit is pointless unless the possibility of mixing has first been eliminated in one way or another.

The ultracentrifuge, first designed by Svedberg for observing sedimentation in highly dispersed sols, is precisely such an apparatus, in which convection has been eliminated and by virtue of this it becomes possible to measure sedimentation accurately at high rotational speeds. An ultracentrifuge differs from ordinary centrifuges in the following basic features: the absence of convection, very powerful and homogeneous force fields, and the presence of an optical device for measurements during rotation. If one adheres strictly to the original meaning of the term “ultracentrifuge,” then of these three features it is precisely the first that is decisive; however, all instruments that serve to obtain high rotational speeds have come to be called by this name, even if these instruments are used for purposes that have nothing in common with sedimentation analysis. Such devices are also included in the present review, especially since they are all based on the same design principles as the so-called analytical ultracentrifuges.

All modern ultracentrifuges may be divided into two principal groups, differing from one another in the idea of their con—

1) McBain showed in his works that the sedimentation of proteins such as egg albumin can be measured at moderate speeds and without an optical device.

structures: the first group includes Svedberg’s ultracentrifuges, the second—a series of instruments developed in American laboratories. In twelve years of work on the ultracentrifuge, Svedberg had transformed it by 1935 into a highly perfected instrument, which is used by him almost without modification even at the present time¹; however, this instrument did not become widespread because of the complexity of its manufacture and its high cost. From 1928 on, in the USA, Beamsᵛ and after him McBain began to develop a new type of ultracentrifuge, based on the principle of an air jet. This direction led to the creation of several types of instruments—from McBain’s simple air-driven tops to the most complex ones. The latest variants, in the perfection of their design, approach Svedberg’s ultracentrifuges. Nevertheless, with respect to the magnitude of working accelerations and the uniformity of the force fields, Svedberg’s machine has remained unsurpassed for precise measurements of sedimentation.

SVEDBERG’S ULTRACENTRIFUGE

As was already indicated above, this centrifuge owes its origin to the fruitlessness of attempts to apply simple centrifugation to the study of sedimentation in colloidal solutions.

While engaged with the question of the size distribution of particles of gold sols, Svedberg passed from observations of simple settling under the action of gravity—possible only for particles larger than 100 mµ—to work with a centrifuge. After the first unsuccessful experiments* he constructed in 1923, during his stay in America, a small centrifuge with an optical device for observations during operation¹. This first instrument still differed in no essential way from ordinary designs. It consisted of a rotor with an electric motor at 20,000 rev/min on a vertical axis with a simple bearing. The rotor consists of two horizontal holders for test tubes, open on the upper and lower sides; both arms are balanced with washers, like the beams of analytical balances. Beneath the rotor a horizontal mirror is fastened, making it possible to take photographs of the test tubes. The centrifugal acceleration was on the order of \(1\,000\,g\). Sedimentation could be observed only for a very short time, at the very beginning of the experiment: the convection currents that subsequently arose disturbed the picture. Having thus established the decisive importance of convection, Svedberg in 1923–1924 systematically worked on the problem of eliminating it; the experiments were carried out with fine-grained gold sols as the test object. In 1924 these experiments ended in complete success.

In the new instrument convection is eliminated in the following way². The cell containing the solution under investigation has the form of a sector with sides directed along the radii of the rotor; the walls do not hinder free settling. The capacity of the cell is \(0.8\ \mathrm{cm}^3\). To reduce external friction and to better equalize temperatures, the rotor rotates in hydrogen at atmospheric pressure. The bearings are cooled with water. The disk-shaped massive rotor, with a radius of 45 mm and a low center of gravity, freely seated—

placed on a vertical axis so that, while rotating, it itself chose the position of equilibrium. In this model the rotation was transmitted to the rotor by means of a gear wheel and a worm screw. The axis of the gear wheel was connected with the drive shaft from the electric motor by a special spring coupling, thanks to which the rotor had free play and was isolated from vibrations of the motor. The guide bearing of the rotor axis was cooled by water, and the gear transmission by oil. The upper and lower walls of the cells were plane-parallel, and a beam directed from below through a small mirror entered the camera. The motor power was 1.5 HP, its speed 5,700 rpm; the rotor speed did not exceed 10,000 rpm. With the aid of this apparatus, which created force fields with accelerations up to 5,000 \(g\), sedimentation was measured and the size distribution of gold sol particles was studied down to the smallest ones. In 1925, using this same apparatus, the first measurements were made of the sedimentation of a protein—hemoglobin. Svedberg called his instrument, which made it possible to separate particles by their size that had previously been inaccessible to investigation, an ultracentrifuge, by analogy with the ultramicroscope and the ultrafilter, which make it possible to observe and separate particles invisible in an ordinary microscope and passing through ordinary filters.

In 1926 Svedberg constructed his first machine of a more powerful type—a high-speed ultracentrifuge, which produced fields of \(100\,000\,g\) at speeds of 45,000 rpm and a rotor radius of 52 mm.\(^{3}\) A massive steel rotor is mounted horizontally on ordinary plain bearings and is driven in rotation by oil turbines. Oil is supplied by a compressor with a 15 HP motor. The bearings, the hollow rotor shaft, and the casing are cooled by a stream of oil. In construction the machine is close to the modern, definitive version; therefore we shall not give a detailed description of it. In the following designs ever-increasing values of acceleration were achieved. In 1931 an acceleration of \(200\,000\,g\) was achieved at a radius of 65 mm and an operating speed of 54,000 rpm. The height of the liquid column was 12 mm, the maximum speed 140,000 rpm. In 1933 the same machine gave an acceleration of \(400\,000\,g\). For a further increase in the intensity of the force field it was necessary to reduce the dimensions of the rotor, thus sacrificing the homogeneity of the force field and the accuracy of the measurements. In 1934 fields of \(900\,000\,g\) were obtained, but at such speeds the rotors burst after several experiments. In 1935 work was constantly conducted with fields of \(130\,000\)—\(710\,000\) rpm.\(^{4}\) In Svedberg’s opinion, a further increase in the intensity of the fields with modern materials is impossible, unless, of course, precise measurements are abandoned.

At present two different instruments are used, selected from a series of experimental designs by Svedberg \(^{\mathrm{I-IV}}\). The first of them is intended for the acceleration range from 500 to 15,000 \(g\) (speeds up to 20,000 rpm), the second for the range 15,000—900,000 \(g\) (speeds up to 150,000 rpm). For speeds of the first range ball bearings are used, which give sufficiently little heating; the required motor power is small. In the design of the high-speed—

...type, sliding bearings with cooling are used. The motors are oil turbines. The rotor must have the mechanically most favorable shape and must be carefully balanced statically and dynamically.

Figure 1 shows a diagram of the apparatus of the first type. The rotor, made of nickel steel, with cell \(C\), is mounted on the vertical axis of a special three-phase motor \(M\); the motor speed is regulated by the frequency of the current within the range from 1,000 to 18,000 rpm. The rotors of the centrifuge and of the motor are enclosed in a chamber through which a slow stream of hydrogen is passed at atmospheric pressure. The motor stator is cooled with water; the rotor chamber is placed in a thermostat \(T\) with a stirrer, cooling, heater, and thermoregulator. The stator is mounted on special rubber supports to absorb vibrations. The cell consists of two round plane-parallel rock-crystal plates 18 mm in diameter and 5 mm thick, which serve as the upper and lower walls.

Fig. 1. Diagram of an apparatus of the first type

Between them is inserted a Bakelite plate with a sector-shaped cutout; the edges of this cutout form the side walls and two cuvettes. All three plates are separated by very thin rubber gaskets (0.06 mm) and inserted into a Bakelite tube. All the parts together are placed in a tube made of duralumin and fastened with threaded rings of the same metal. The cell is inserted into a round socket of the rotor. Evaporation of the liquid in it is prevented by applying a thin layer of paraffin to the surface. A beam of light from the mercury lamp \(L\) passes through light filters \(F_1, F_2, F_3\), is reflected by prism \(P\) into the cell, and falls into the camera \(K\) with the long-focus objective \(O\). \(S\) is an electromagnetic shutter. Photographs of the solution taken during rotation (in ultraviolet light) are photometered, and the curves are used for subsequent calculations.

Fig. 2. Diagram of apparatus of the second type

We present the diagram of the device of the second type in Fig. 2.

The rotor, made of chromium-nickel steel, rests on horizontal bearings \(B_1\) and \(B_2\); the motors are two small oil turbines \(T_1\) and \(T_2\) at the ends of the rotor shaft. Hydrogen is passed through the space surrounding the rotor, while continuous evacuation...

its pressure is maintained at approximately 25 mm. At this pressure the thermal conductivity of hydrogen is not yet reduced, while the friction is insignificant. An important role is played not only by the pressure, but also by the purity of the hydrogen: an admixture of several percent of air already produces convection currents in the cell. The temperature in the bearing and in the rotor chamber is measured by thermocouples \(Th_1\) and \(Th_2\); the temperature of the cell at various rotational speeds is determined by observing the melting of preselected low-melting-point alloys, which for this purpose are placed in the vessel. The light beam from the mercury lamp \(L\) passes through the filters \(F_1\), \(F_2\), \(F_3\), the rotor cell \(C\), and enters the photographic camera \(K\). Photographs are taken with the aid of two electromagnetic shutters \(S_1\) and \(S_2\). The number of revolutions is determined stroboscopically. Oil is supplied to the turbines by a specially designed compressor and is precooled. The oil for lubricating the bearings is supplied by the same compressor, is carefully filtered, and its pressure is set by means of the valve \(V_2\) equal to \(2\ \mathrm{kg/cm^2}\). The oil pressure in the turbines, on which the number of revolutions depends, can be varied within the limits from 0 to \(19\ \mathrm{kg/cm^2}\) by means of the valve \(V_1\). Resistance thermometers \(R_1\), \(R_2\), \(R_3\) and manometers \(M_1\), \(M_2\), and \(M_3\) serve for measuring the temperature and pressure in the various parts of the apparatus. To prevent accidents in the event of rotor rupture, the upper and lower halves of the massive steel stator are fastened with screws of chromium-nickel steel, which are embedded in a concrete foundation.

Fig. 3. Rotor of the second type

The rotor of the first 1933 design (Fig. 3) has an oval shape with a greatest diameter of 180 mm. Its thickness at the periphery is 52 mm, its weight 8.9 kg. The distance of the centers of the openings for the cells from the rotor axis (the effective radius) is 65 mm. In the peripheral part, between the openings, a considerable amount of material has been removed to lighten the rotor; only thin ribs have been left. The outside diameter of the cell is 26 mm, its weight 56 g. The height of the liquid column is 12 mm. At a speed of this rotor of \(78\,000\ \mathrm{rev/min}\), a residual elongation appeared in the direction of the line connecting the cell seats. The working speed is \(75\,000\ \mathrm{rev/min}\), which corresponds to \(400\,000\,g\) at the centers of the cells; for the given dimensions of the cells and rotor this is a limit that can be raised only slightly by replacing duralumin with a magnesium alloy.

For a further increase in the intensity of the force field it is necessary to reduce the diameters of the rotor and of the cells; however, the requirements of measurement accuracy set a limit to such reduction. These requirements are as follows: the inhomogeneity of the force field in the cell must not exceed \(25\%\); the minimum permissible height of the column ...

liquid is 6 mm\(^{1}\). The first condition determines the distance of the cell socket from the axis—about 36 mm (with a large rotor diameter of 102 mm). The second condition determines the external diameter of the cell—16 mm. Such a rotor, in 1933, made it possible to obtain fields of \(600\,000\,g\). With an extreme reduction of the effective radius to 32.5 mm in 1935, as already indicated above, accelerations of \(700\,000\,g\) were obtained. However, Svedberg notes that the accuracy of measurements with the rotor of the second design is unsatisfactory.

Measurements of sedimentation make it possible to determine the molecular weight of a dissolved substance in two ways: the first, used in relatively small force fields, is the observation of the so-called “sedimentation equilibrium,” which occurs when the rates of the sedimentation and diffusion processes are equal; the second method consists in measuring the rate of sedimentation, which is possible only in large force fields. The rate of sedimentation, referred to a unit force field and to water at 20° as the solvent, gives the molecular sedimentation constant \(s\):

Fig. 4a. Photographs of a column of Limulus hemocyanin solution in a cell

Fig. 4b. Sedimentation curves of a Limulus hemocyanin solution

\[ s=\frac{dx}{dt}\cdot \frac{1}{\omega^{2}x}\cdot \frac{\eta}{\eta_{0}}\frac{1-V\rho_{0}}{1-V\rho}, \]

where \(V\) is the specific volume of the dissolved substance, \(\frac{dx}{dt}\) is the sedimentation velocity, \(\eta\) is the viscosity of the solvent, \(\rho\) is the density of the solvent, and \(\eta_{0}\) and \(\rho_{0}\) are the corresponding quantities for water at 20°. The difference in sedimentation constants determines the separation of the dissolved molecular mixture.

In Fig. 4a a series of photographs of the sedimentation of Limulus hemocyanin is shown. The usual intervals between photographs are 5–10 min. Analysis reveals the presence of molecules with four different molecular weights. In Fig. 4b the sedimentation curves of the same solution are given. With the aid of an ultracentrifuge it is possible to determine molecu-

\(^{1}\) The “separating capacity” of a centrifuge with respect to the components of a molecular mixture is proportional to \(\omega^{2}XH\), where \(\omega\) is the angular velocity, \(X\) is the effective radius, and \(H\) is the height of the solution column.

molecular weights of high-molecular substances of low density (\(M\)—from 6,000,000 to 1,000), as well as the molecular weights of low-molecular substances with high densities (\(M\) up to 200).

ULTRACENTRIFUGES BASED ON THE PRINCIPLE OF THE AIR JET

During the first years after the appearance of Svedberg’s high-speed ultracentrifuge, he remained its sole owner. The need for especially precise balancing and centering of the rotor, the presence of special compressors, and a complicated cooling system—all this made the machine inaccessible to other laboratories. It is natural, therefore, that the efforts of designers turned toward a radical change in the principle of construction, and above all toward such an improvement of the supporting devices as would make it possible to attain high speeds without appreciable friction and heating and with a minimum expenditure of energy.

The way was indicated in 1925 by Henriot and Huguenard, who caused a small rotor to rotate in an air jet without mechanical contact with the stator \(^5\). It is known that a jet of air emerging from a cone-shaped recess draws in a small ball brought close to it and attracts it toward the funnel up to such a distance from its walls at which this attractive action of the air current and the weight of the ball are balanced by the pressure of the air on the ball from below. The air top of Henriot and Huguenard is based on this principle. It consists of a small cylindrical rotor with a conical lower surface, on which there are radial grooves; the stator is a metal funnel, to the walls of which straight tubes are brought up, arranged like the generators of a one-sheeted hyperboloid of revolution. Their openings on the inner surface of the funnel lie at one level—in a circle. Compressed air, issuing from the tubes, forms a vortex air stream. The cone-shaped rotor is drawn in by this vortex, is balanced at a distance of tenths of a millimeter from the walls of the funnel, and begins to rotate at high speed. The rotor is not balanced and itself selects the position of the axis of rotation. As the air pressure is increased, the speed increases; the authors emphasize that its limit is set only by the mechanical strength of the rotor. At an air pressure of \(3\ \text{kg}/\text{cm}^2\), the number of revolutions of one specimen reached 11,000 rev/sec; copper rings placed on the rotor burst. The rotors were made of strong steel. The experimenters attempted to use the air top as a centrifuge; however, its stability still left much to be desired. During rotation the rotor vibrated, emitting an audible tone of definite pitch. At very high speeds it jumped out of the stator.

Following Henriot and Huguenard, the air top was studied in detail, chiefly in the USA \(^6, ^7, ^{VI}\): the conditions of its stability were investigated, as was the aerodynamically most advantageous shape of the longitudinal section of the supply channels, the dependence of the speed on their number, and the optimal angles of their axes with the horizontal plane and the angles

horizontal projections of the axes with the radii of the stator. For the highest speeds the former are equal to 35°, the latter to 65°. It turned out that the stability of the rotor can be increased by simple means; one such means is free (or regulated) access of air to the lower part of the stator cone through an opening at the lower point—the apex of this cone. In this case the vertical vibration ceases, which is caused by the rarefaction of air formed between the stator and the rotor at high speeds. The rotation of the rotor then becomes so stable that its load can be sharply changed at full speed (for example, by pouring liquid into it) without disturbing its equilibrium. The rotor can rotate a system that is not very carefully balanced. The air top is very sensitive to all kinds of resonance vibrations in the stator; to absorb them, the stator is mounted on rubber gaskets. The occurrence of vibrations in the system supplying the air—in the valves and supply tubes—should be avoided, since such vibrations are transmitted to the rotor through the air cushion. Nonperiodic sharp lateral displacements of the rotor are also observed; they are caused in part by its imperfect dynamic balancing and by eccentricity of the axis of rotation—gradually increasing, this eccentricity can lead to contact of the rotor with the stator. Therefore it is best to mount the stator so movably that it follows the horizontal displacements of the rotor; for this, in addition to flexibility of the mounting, the mass of the stator must be less than the mass of the rotor. With the aid of such a device, Girard and Suckrie achieved excellent stability of the rotors, obtaining speeds of 4,000 rev/sec and 6,000 rev/sec with corresponding rotor diameters of 20 and 13 mm. The centrifugal accelerations at the periphery of the rotor were from 600,000 to 1,000,000 \(g\). With increasing rotor size, friction against the air rapidly increases, and consequently so does the power required for high rotational speeds. Curves of the dependence of rotor speed in air on the pressure of the moving air are given in Fig. 5.

Fig. 5. Curves \(A\), \(B\), \(C\), and \(D\) were obtained with an immobile stator and with the rotors shown to the right of the curves. Curve \(E\) was taken with a movably mounted stator (Beams)

If the moving air is replaced by hydrogen, the friction decreases, and the velocity of the gas streams increases. Beams\(^{VI}\), with a rotor 9 mm in diameter in a stream of hydrogen, reached the record rotational speed reported in the literature—21,600 rev/sec, which corresponds to a centrifugal

acceleration exceeding \(8{,}000{,}000\,g\). By rotating small mirrors with the aid of the rotor, Beams took photographs of electric sparks; in doing so he succeeded in registering time intervals of the order of \(10^{-9}\) sec.

Further improvement of the air top, with the aim of turning it into an instrument for quantitative measurements, proceeded in two directions. MacBain tried to increase the smoothness of the rotor’s motion and to eliminate convection by comparatively simple means, while preserving the general scheme of the air top unchanged; the second direction, represented by Beams and his collaborators, created on the basis of the air jet an entirely new, considerably more complex design.

MACBAIN’S ULTRACENTRIFUGE

A detailed description of one of MacBain’s instruments appeared in 1935.^8 Its drawing is given in Fig. 6.

MacBain discovered yet another cause of horizontal displacements of the rotor: turbulence of the air stream after it leaves the tubes. To avoid the influence of vortices, he fixed on the stator three vertical, radially directed partitions. Between them and the rotor there remains a gap of \(1\) mm, sufficient to ensure that the partitions do not affect the speed of the rotor. Thanks to this device, the horizontal trembling at moderate air pressure is imperceptible under the microscope; at high pressures of the moving air it has an amplitude of \(0.01\) mm. Incorrect positioning of the stator and clogging of the air-supply tubes cause precession of the rotor. Therefore the stator is set exactly level; the compressed air is passed through a fine mesh. The other conditions for stability of the air top have already been mentioned; they are likewise provided for in the design of the stator. A short glass tube, attached to the stator by means of a piece of rubber hose, serves for the central supply of air. A long tube or hose may be the cause of vibration. The admission of air is regulated by a screw that compresses the hose. The chamber supplying the moving air is mounted on a very long metal feed tube. This tube is led vertically and below is brought out to the central axis of the instrument; thanks to this the stator system has great flexibility and mobility in the horizontal plane. The chamber is surrounded by a lining of rubber sponge. To eliminate convection, MacBain regulates the temperature of the air surrounding the rotor. If the air in the chamber remains at room temperature, then at the top of the rotor the temperature will prove to be \(5^\circ\) higher than in the cone cavity, while the temperature of the stator will be \(2^\circ\) lower than in the cavity: the air jets issuing from the orifice are cooled. Under such conditions sedimentation would be impossible; therefore the driving air is passed through a large copper coil immersed in a heated water thermostat. The rotor is protected from above against currents of room air by a metal cover, insulated with asbestos, which is placed on the casing. The high-pressure supply is thermally insulated from the casing

and the frame by means of a bakelite bushing. The entire installation is enclosed up to the very top by a thick case stuffed with wool. The air for the central feed is sucked in from the housing. By regulating the temperature of the thermostat and the magnitude of the driving pressure, it is possible to achieve complete equality of the temperatures above and below the rotor and to maintain this equality with an accuracy exceeding \(0^\circ,1\). The temperature of the cell walls is determined approximately by means of two thermocouples: one junction is fastened directly above the center of the outer surface of the rotor, \(0.5\) mm above it; the second junction is introduced into the hollow cone of the rotor as close as possible to the center of the lower surface; the third thermocouple is introduced into the stator.

The rotors consist of two screw-connected parts—the upper cylindrical part, accommodating the cell for the liquid, and the lower part, having the shape of a truncated cone, which in fact serves as the air turbine. The lower part is hollow, and in its bottom there is an opening \(13\) mm in diameter. Through this opening, during rotation of the rotor, a stationary periscope with a quartz prism at the end enters the cavity. The light beam, traveling along the axis of the stator, is directed by this prism into the cell. The light of a high-pressure mercury arc passes through a monochromator; otherwise the optical arrangement does not differ from Svedberg’s. The clearance between the head of the periscope and the cone of the rotor is \(2\) mm. The periscope tube also serves simultaneously for the central air feed.

Two types of upper rotor cylinders carrying the cells have been developed. The first of them is a steel frame screwed onto the lower part of the rotor; into it are inserted two quartz disks \(30\) mm in diameter and \(2\) mm thick, forming the upper and lower walls of the cell. The disks are separated by a gasket which serves as the bottom of the cell. The gasket consists of a rubber ring (outer diameter \(27\) mm, inner diameter \(15\) mm, thickness \(1.1\) mm); on the outer edge of the ring a thin strip of rubber (from a rubber glove), \(5\) mm wide, is placed. The rubber ring, in which four radial cutouts have been made, is inserted into a brass ring directly adjacent to the wall of the rotor. During rotation of the cell in the centrifuge with several drops of liquid, the rubber strip is pressed against the quartz and brass and hermetically closes the gaps; the tightness improves as the speed increases. The quartz disks are closed on the outside by two steel disks \(30\) mm in diameter and \(2\) mm thick: one disk above and the other below. Each of them has four round openings \(6\) mm in diameter, at a distance of \(9.5\) mm from the center of the disk; these openings lie opposite the four cutouts in the rubber ring and form the windows of the cell. The lower steel disk is inserted into the rotor by means of two pins; all the parts are screwed together. Finally, in order to avoid friction of the steel disks with the window against air, the upper one is closed by a quartz disk, and the lower one by a small circle of celluloid or cellophane. Thanks to the steel disks, the limiting speed that the quartz walls can withstand can be increased by \(30\%\), and the centrifugal acceleration correspondingly by more than \(70\%\). If the entire surface of the disks is optically

polished, they withstood a speed of 3,000 rev/sec (540,000 g). However, the slightest scratches on the quartz can greatly reduce its strength: a disk that withstood speeds of 2,500 rev/sec unexpectedly broke at 1,700 rev/sec. Therefore, the limiting speed for these cells should be taken as 1,500 rev/sec. Another type of device for cells is shown in Fig. 6. The upper part of the rotor is a steel cylinder 8.2 mm high, with two holes 8.6 mm in diameter, located at a distance of 10 mm from the center of the rotor. Two tubes of Monel metal are inserted in them: one of them is tightly closed and serves for balancing, the other carries the cell. The latter consists of seven parts: a rubber ring with a cut, two thin rubber strips, two quartz windows, a brass washer, and a hollow brass screw; the radial cut has an angle of 6° and a length of 4 mm. The upper window is conical in shape, 2 mm thick. The outer surfaces of the rotor, the tube, and the quartz window are precisely fitted and form one plane. During assembly the cell is compressed from the outside, and under pressure it is held by the hollow screw. The liquid is introduced into it through a small lateral opening and covered with pure paraffin oil. This type of rotor, lower than that described at the beginning and, owing to the reduction of friction against air, gives speeds greater by 17%. No ruptures of the quartz windows occurred.

Fig. 6. Details of the rotor of the second type, stator and periscope.

A—cell holder, B—rotor cone, C—blank cell, D—cell, E—quartz disks, F—rubber gasket, G—periscope prism, H—periscope tube, J—stator, K—bronze inserts, L—disk on which reinforcing partitions are mounted (not shown in the drawing), M—high-pressure chamber, N—hermetic nut, P—lock nuts, R—holder of the central air supply, S—central-supply tube, T—window, V—high-pressure supply.

The material used for the rotors was chrome-manganese steel. Duralumin proved less suitable than steel: with strength more than six times lower, its density is 0.354 that of steel. The rotational speed of the rotor at constant diameter depends on the velocity of the outflow of air from the openings; the latter, as the air pressure is increased, rapidly tends to a limit. The curve of the dependence of the outflow velocity on pressure shows that it is not advisable to use pressures exceeding 14 kg/cm². Therefore everything possible was done to increase the rotor speed at constant pressure. The speed can be increased by reducing the pressure in the space surrounding the rotor; however, this increases vertical vibration. For the cone angles of the stator and of the rotor, the corresponding values of 100° and 90° were chosen.

To minimize surface friction, the rotors must be as low as possible and must have a well-polished surface. The maximum speed achieved by Mac Bain was

for a rotor of diameter 37 mm is equal to 3,000 rev/sec. The speed is kept constant with fluctuations of less than 0.1%. Compressed air from the compressor is supplied through two cylinders with one reducer between them and one at the outlet. Measurements of the speed are made by comparing the sound tone of the rotor with the beat tone of a calibrated circuit. Another method consists in obtaining a pulsating current from a photoelement on which light falls after being reflected by the lateral surface of the rotor, half blackened; the frequency of the pulsating current is compared in a telephone with a known beat frequency.

The centrifuge described made it possible to measure the sedimentation constants of proteins. However, good agreement with Svedberg’s results was not obtained. The mean value for egg albumin obtained with the McBain–O’Sullivan rotor is \(2.98 \cdot 10^{-13}\), whereas according to Svedberg it is \(3.55 \cdot 10^{-13}\). Apparently, despite all efforts, it was not possible completely to eliminate convection currents. McBain therefore developed a number of methods for mechanical retardation of mixing \(^{9,10}\). In 1938 he described one such construction, which proved to be the most successful \(^{10}\). In it a closed-type rotor is used (Fig. 7,a).

Fig. 7. Section of a closed-type rotor (a) and an insert for it (b)

\(A\)—cover, \(B\)—cone, \(C\)—cellophane sheet, \(D\)—metal circle, \(E\)—metal washer

The sedimentation rate is determined analytically; no optical devices are required. The upper part \(A\) has an outside diameter of 32.5 mm and a height of 11.5 mm. The internal diameter of the recess is 28 mm, the height 9 mm. Into this recess is inserted a stack of horizontal circular plates joined to one another (Fig. 7,b).

Thus the whole cell is divided into a series of compartments communicating with one another. The distance between the partitions is 0.08 mm. They do not hinder sedimentation and do not affect its rate. The material for the insert is coin silver, stainless steel, or iridium-platinum. The diameter of the large disks is 24.6 mm, thickness 0.102 mm. The diameter of the smaller disks is 13.6 mm, thickness 0.076 mm. The insert is fastened by two side disks and a vertical pin. The sedimentation rate is determined by analyzing samples of liquid taken from the cell with a syringe. For this purpose there are openings (of \(1 \text{ mm}^2\)) in the upper and lower side disks. Samples are first taken from the “external,” unpartitioned volume, which is then emptied; the “internal” liquid then capillarily

is held between the plates of the insert. The rotor is then run for a short time; the liquid remaining in the compartments is squeezed out of them, and a sample can be taken to determine the concentration \(c\). The method consists in measuring the rate of advance of the sedimentation boundary between the levels of the edges of the large and small disks; the position of this boundary is given by the formula

\[ x=\sqrt{\frac{b^{2}H}{(P+H)}} , \]

where \(x\) is the position of the boundary;

\[ H=\frac{a^{2}}{(b^{2}-a^{2})}, \]

where \(a\) and \(b\) are the radii of the disks;

\[ P=\frac{c}{c_{0}}, \]

where \(c\) is the measured concentration of the liquid held between the disks, and \(c_{0}\) is the initial concentration of the liquid between \(a\) and \(b\). The number of revolutions of the rotor reaches 2,000 rev/sec. The mean value obtained for the sedimentation constant of egg albumin is \(3.56\cdot10^{-13}\), while Svedberg’s latest result is \(s=3.55\cdot10^{-13}\). The result for the air-driven turbine is remarkable, if one takes into account the absence of an optical device and the extreme simplicity of the construction.

Fig. 8. Vacuum ultracentrifuge with a rotor for isotope separation

The same designations are retained for Fig. 10

THE BEAMS ULTRACENTRIFUGE

In order to obtain, with the aid of an air-driven turbine, more homogeneous force fields, it is necessary to increase considerably the dimensions of the rotors. As their diameter increases, air friction grows rapidly and, consequently, so does the power required for rotation, while the limiting rotation speed decreases. For a rotor 200 mm in diameter the limiting speed is 16,000 rev/min. Beams decided to place the rotor of the air-driven turbine in a vacuum; this served as the impetus for the creation of a new design of air ultracentrifuge V, VI, 11, 12.

The main features of this design are as follows. The moving part of the rotor is transformed into an air turbine and separated from the “working” part, the rotor proper. The latter is placed in a vacuum suction chamber.

chamber and connected to the turbine by a flexible shaft. The flexible shaft passes through a gland that hermetically seals the chamber. Finally, the supporting stream of air is separated from the driving streams and is regulated independently of them; it creates, under the turbine, an air cushion supporting the entire rotating part of the apparatus. The scheme of one of Beams’s designs VI is shown in Fig. 8 (see also the photographs in Figs. 9 and 10).

The rotating parts consist of the turbine \(T\), the flexible shaft \(A\), and the “centrifuge,” or rotor \(C\). Initially, piano wire was used for the shaft. In the present design it is a flexible tube passing through two hermetic glands \(G_1\) and \(G_2\). \(G_1\) seals the vacuum chamber; \(G_2\) makes it possible, while the rotor is rotating, to fill it with liquid or gas, or to evacuate it.

Fig. 9. Photograph of the apparatus shown in Fig. 8, with the vacuum chamber removed

Fig. 10. Photograph of the turbine, flexible rod, air-support ring, neoprene rings, sealing glands, and upper cover of the chamber

To make the gland, a brass rod is turned, and a channel is drilled along its axis. Into this channel, plugs of soft metal are inserted from both ends; channels are also drilled in them, of somewhat smaller diameter than \(A\), and they are bored out just enough so that the rod slides in them without effort. If the rod is steel, the plugs may be made of brass, bronze, or babbitt. \(G_1\) and \(G_2\) are mounted on round elastic rings \(R\) of neoprene or another material not affected by mineral oil. In this way the glands have a certain mobility. Oil of low vapor pressure (for vacuum pumps) is forced into them through the channels visible in the drawing. Through such a gland, less than \(1\ \mathrm{cm}^3\) of oil seeps into the evacuated chamber during 1 hour of operation of the apparatus. The device provides sufficient heat transfer from the gland to

to the cover of the vacuum chamber \(P\). With proper operation, Babbitt plugs usually need to be replaced only after \(10^9\) revolutions of the rotor. The shaft \(A\) is made of steel, copper-beryllium alloy, or another strong metal. The diameter of the turbine depends on the weight and rotational speed of the rotor \(C\); it is made of duralumin, phosphor bronze, or steel. The air bearing is formed in the space between the Bakelite ring \(B\) and the lower surface of the turbine \(T\). If the design does not provide for a second oil seal \(G_2\), then this surface is made conical. The stator is arranged in the same way as in a simple air turbine.

The increase in the load of the centrifuge raises the friction in the oil seals and the friction against the air so insignificantly that it almost does not require an increase in turbine power. Therefore the maximum speed, as in Svedberg’s apparatus, is determined solely by the strength of the rotor. Special steels serve as the material for the rotor; certain grades of duralumin are also used. The vacuum chamber must withstand ruptures of the rotors—it is a steel cylinder and plates at least \(2.5\ \mathrm{cm}\) thick. The chamber must be carefully thermostated.

Before the apparatus is started, the rotor \(C\) is fixed in the required position on the shaft \(A\). Then air is admitted into channel \(I\), and the pressure (it is of the order of \(0.45\ \mathrm{kg}/\mathrm{cm}^2\)) is increased until the moving parts are lifted by the air bearing. At approximately the same pressure, oil is supplied to the oil seals; the chamber is evacuated. The magnitude of the vacuum is determined by the vapor pressure of the oil and reaches \(10^{-6}\)—\(10^{-7}\ \mathrm{mm}\ \mathrm{Hg}\). Then the turbine is started. During the increase in speed, vibrations of the rotor are observed corresponding to the frequencies of elastic oscillations of the apparatus itself, the chamber, the shaft \(A\), etc. The operating speed must be sufficiently far removed from these critical numbers of revolutions. Outside them, the apparatus works so smoothly that the rotation is difficult to notice from the outside. The operating speed is \(1550\ \mathrm{rev}/\mathrm{s}\) and is maintained by a pressure of \(0.7\ \mathrm{kg}/\mathrm{cm}^2\).

At a speed of \(1650\ \mathrm{rev}/\mathrm{s}\), destructive deformations of the rotor appear. To stop the apparatus, braking, oppositely directed air jets are used. The speed is measured either stroboscopically or with the aid of electromagnetic devices \(^{13,14,15}\). The simplest of them is as follows \(^{13}\).

The measuring device consists of a coil, the ends of whose winding are connected to a Wheatstone bridge. A small magnet attached to the centrifuge rotor, when rotating, induces a current in the coil; the frequency of this current is measured by the bridge. In one of the branches of the latter, in addition to the ohmic resistance \(R_1\), the capacitance \(C\) and inductive resistance \(L\) are included. The bridge can be balanced only at the frequency

\[ N=\frac{1}{2\pi\sqrt{LC}}. \]

By changing \(L\) or \(C\) until there is no current in the null instrument, one can calculate \(N\). A piece of sewing needle \(1\)—\(2\ \mathrm{cm}\) long served as the magnet. The resistances of the remaining branches of the bridge were \(R_2=R_4=300\ \Omega\); \(R_3=100\ \Omega\); \(R_1+L=100\ \Omega\); \(C\)—mica capacitors, \(L\)—variable self-inductance. Similar in prin-

DESIGNS OF MODERN ULTRACENTRIFUGES

A simple device was also proposed for automatic speed regulation[^14].

A magnetized steel bar, positioned along a diameter, is fastened to the rotor. On a special stator there is a two-pole winding included in a circuit with a natural oscillation frequency corresponding to the required number of revolutions of the rotor. After the air turbine is started, the rotor tends to rotate synchronously with the oscillations excited in the winding, and resists any further increase in speed.

At a natural frequency of 1,000 cycles, the driving air pressure for a speed of 1,000 rev/sec was 2.8 kg/cm²; to increase the speed by 2% a pressure of 6.3 kg/cm² was required. By feeding the circuit with current from outside, the rotor can be made to rotate synchronously by an electromagnetic method. This was used by Beams and Snoddy for an “electrovacuum” centrifuge (see below).

The apparatus described above was used for separating isotopes by fractionating carbon tetrachloride[^5][^6][^17][^18].

Into a rotor of 65 cm³ capacity, 15 cm³ of CCl₄ is poured; the apparatus is then set in operation and rotor C is evacuated through the hollow rod A. CCl₄ evaporates at the periphery of the receiver, diffuses toward the center against the centrifugal force, exits through A, and is collected in dry-ice traps in three equal fractions. After thorough purification, the fractions show differences in density approximately corresponding to those expected on the basis of theoretical calculations for isotope separation in a centrifugal-force field. In this experiment the chamber is not pumped out completely—such an amount of air is left in it (10⁻³ mm Hg) that, during friction, the heat necessary for evaporation of CCl₄ is developed.

This centrifuge gave good results with respect to agreement of the fraction yields with theoretical data; however, its drawback is low productivity: the amount of substance that can be centrifuged simultaneously is limited by the capacity of the rotor. To increase the “throughput,” Beams constructed a tubular ultracentrifuge[^5][^16] (Fig. 11).

The rotating parts of the apparatus are: turbine T, tubular rotor R, and flexible hollow rods S₁, S₂, and S₃. S₁ passes through the oil-sealing gland G₁. S₂ passes through glands G₂ and G₄, and S₃ through G₃. G₁ and G₂ seal the vacuum chamber V; G₃ and G₄ make it possible to separate the light fraction, issuing through S₂, from the heavier fraction, which exits between the coaxial tubes S₂ and S₃. Gland G₁ is fastened on the metal disk N, the lower surface of which is ground to the cover of the vacuum chamber. Through the tubes O, oil is delivered under such pressure that disk N rises slightly and “floats” on a thin layer of oil. Such a sealing device is necessary in order to give the shaft sufficient mobility. The gas being centrifuged enters the rotor R continuously through channel A; as it moves downward, separation takes place. The light fraction is directed downward through S₂ and is collected in chamber C; the heavy fraction flows out through

the gap between the inner wall of the rotor and the circular plate of the bottom, and then through channels \(H\) into the space between \(S_2\) and \(S_3\), and is collected in chamber \(B\).

The ratio of the amounts of substance collected in \(B\) and \(C\) can be regulated by the corresponding valves. For operation with liquids, a solid steel cylinder is placed inside the rotor tube, with a radius smaller than the rotor radius. The substance is thereby displaced into the region with the greater magnitude of centrifugal force, and the rate of separation increases. The height of the rotor is not limited by the stresses arising during rotation; therefore the capacity, and consequently the productivity, of such a centrifuge can be extremely large. The apparatus was used for separating gases, mixtures with a constant boiling point, and high-molecular substances in solutions. Carbon tetrachloride vapor was admitted at a rate corresponding to \(2\ \text{cm}^3\) of liquid \(CCl_4\) per hour. However, the yield, unlike the productivity, proved worse than in rotors of the first type—\(40\%\) of the theoretical value.

Fig. 11. Diagram of a tubular centrifuge

Beams also carried out experiments on separating gases by a third method—according to the principle of velocity selection \(^{1,17,18}\). A rotor with a triangular horizontal cross-section had openings on the periphery arranged so that only those molecules whose velocity exceeded the linear velocity of the periphery could enter them. The gas that had entered the rotor left through radial channels to the hollow shaft and was pumped out. The peripheral velocity at ordinary temperatures considerably exceeds the mean molecular velocity, and thus, along with centrifugation, separation of molecules according to velocities also occurs.

Experiments on separating a gas mixture of \(N_2\) and \(CO_2\) gave good agreement with theory. The separation factor is equal to

\[ \frac{k_0}{k}=e^{\left(\frac{v^2}{RT}\right)(M_2-M_1)}, \]

where \(k_0\) is the ratio of the amount of light isotopes to heavy ones at the center of the rotor; \(k\) is the same at the periphery; \(v\) is the peripheral velocity, equal to \(8\cdot 10^4\ \text{cm/sec}\); \(M_1\) and \(M_2\) are the molecular weights of the light and heavy isotopes, respectively. For \(T=80^\circ\) and \(M_2-M_1=4\), the value of \(\frac{k_0}{k}=6.9\). At this temperature the elasticity of the vapors of many substances is already sufficient for centrifugation in the gas phase; therefore a rotor has been designed that operates at the temperature of liquid air \(^{18}\).

The analytical model of Beams’s ultracentrifuge was constructed by Bauer and Pickels[^19][^20]. The form of the rotor for quantitative measurements is shown in Fig. 12. Among the improvements in the design one may note the movably mounted air-bearing ring, which absorbs vibrations; a device for collecting oil seeping into the chamber; and an ingenious method of measuring the temperature of the rotor by means of a small cup with mercury, into which, during rotation, the steel tip of the rotor is immersed. The cup is made of ebonite, is movably suspended on silk threads, and is closed on top so that the mercury does not splash out of it during rotation.

The wall of the chamber is a cylinder of chrome-nickel steel 7.6 cm thick. The oval-shaped rotor, with a major diameter of 185 mm, weighs 3.430 kg. A Svedberg-type cell is located at a distance of 65 mm from the rotor axis and is designed for a liquid column 15 mm high. The operating speed is 60,000 rpm; the acceleration in the cell is 260,000 g. The pressure of the driving air is 1.25 kg/cm². The rotor material is either alloy steel or a special aluminum alloy. The chamber is evacuated to \(10^{-3}\) mm Hg. After several hours of operation the temperature of the rotor rises by \(1\)—\(2^\circ\). At the same time, no convection currents that would disturb sedimentation are observed. At operating speed the centrifuge rotates so smoothly that, in a cup placed on the plate carrying the seal, the surface of the mercury remains perfectly calm. The centrifuge is stopped by braking reverse air jets. The optical device does not differ from Svedberg’s.

Fig. 12. Rotor of the “analytical” type—Bauer and Pickels

Beams developed a number of variants of the basic design: an “inverted” centrifuge, in which the rotor is mounted on a flexible rod above the turbine and has great mobility[^6]; a tubular-centrifuge-type device for rotating long bars and rods with two flexible shafts[^6]; a centrifuge with horizontal bearings and a flexible shaft[^12]. In the original designs there was a stabilizer under the rotor, which fixed its position during rotation[^12].

Coakston and Beams described a centrifuge with a steam engine[^21]. Continuing to improve the centrifuge mechanism, Beams and Snoddy replaced the air motor with an electric one[^22][^23]. The place of the turbine in this instrument is occupied by the rotor of a “squirrel-cage” type electric motor; it is set in motion by current in the stator windings. The air bearing is arranged, as usual, under the rotor. The motor is powered by an audio-frequency amplifier with a maximum power of 300 W. The frequency is produced by the beats of two piezoelectric generators.

The advantage of the electric motor consists in the automatic setting of the rotor at a constant speed. After starting, the number of revolutions of the centrifuge gradually increases until it becomes practically equal to the frequency of the alternating current in the windings. This equality is achieved because the amount of friction is negligible and is reduced to the friction of the rotor against air. In this case, the degree of constancy of the speed corresponds to the constancy of the frequency of the circuit. Heating of the moving rotor does not affect the temperature of the vacuum chamber, which is regulated in the usual way.

Various circuits of the supply circuits and other types of motors were used, in particular a synchronous motor, which gave 1,400 rev/sec. For speeds of the order of 1,000 rev/sec the electric motor does not require a complicated arrangement; it eliminates the need for a high-pressure air supply and can operate without supervision.

In a review of his worksv, made at the end of 1938, Beams reported a new type of support for the rotating parts of the ultracentrifuge. The flexible shaft of the instrument is lengthened and projects above the upper gland. At the end of it a short steel cylinder is fastened to the rod in such a way that the rod enters a narrow channel drilled along the axis of the cylinder and projects slightly above the upper plane of the latter. Directly above the end of the rod, and coaxially with it, a solenoid is installed, which attracts the steel cylinder and in this way supports the entire rotating part of the instrument. The end of the rod enters a small recess on the lower side of the solenoid and rests against the bottom of this recess; at the same time a narrow gap remains between the solenoid and the cylinder. The current in the solenoid is calculated so that the cylinder is attracted with a force slightly exceeding the total weight of the rotating parts; therefore the pressure of the end of the rod on the support is very small. When the rotor rotates, the friction of the lubricated rod in such a thrust bearing is less than the friction in the glands of the instrument. Magnetic supports were used to support rotors weighing over 1 kg and, apparently, are also suitable for considerably heavier rotors. The rotational speeds obtained were the same as with an air support; no appreciable braking due to Foucault currents was observed. Thus the air ultracentrifuge was converted into an electric one: the combination of magnetic support and electric motor makes it possible to dispense with compressed air. The current is taken from a 110 V mains. To rotate rotors 15 cm in diameter at speeds above 1,000 rev/sec, or 10 cm in diameter at speeds up to 1,200 rev/sec, only 275 W were required. During 6 hours of operation without regulation the centrifuge showed no fluctuations in speed.

An even more perfected type of magnetic suspension was realized by Holmes24, 25. His device has not yet found application in working models of ultracentrifuges. However, the first attempts in this direction have been made; moreover, Holmes’s suspension deserves attention also because it is the successful completion of a series of efforts by experimenters to reduce to a minimum the retarding moment of friction. The device consists of a magnetic needle placed in vacuum, hanging vertically without mechanical support in the field of a solenoid.

The needle is in a position of stable equilibrium—in the position with minimum potential energy, in which the force of gravity and the force of the magnetic field are equal and oppositely directed. The magnetic field is produced by two solenoids arranged coaxially, one above the other. The first, upper solenoid is supplied with direct current so as almost to balance the needle with the cylinder fastened to it. The upper edge of the cylinder half obstructs the beam of light incident on the photocell, which regulates the current of the amplifier feeding the second solenoid. Thus equilibrium is established automatically: the field strength is weakened when the needle is too high and increased when it is too low. Damping of the vertical oscillations of the needle was carried out electromagnetically, for example by introducing the appropriate self-inductance or capacitance into the supply circuit of the second solenoid.

In the experiment on rotation, a strong disk of aluminum alloy, 15 mm in diameter and 5 mm thick, was attached to a magnetic rod 55 mm long. The disk together with the needle weighed about 6 g and had a moment of inertia of \(0.8\ \text{g}/\text{cm}^{2}\). It rotated in an evacuated chamber at a pressure of \(10^{-5}\ \text{mm Hg}\) under the action of the field of a small magnetic bar, which was fixed horizontally beneath the disk outside the chamber on an air turbine rotating at a speed of about \(1\,500\ \text{rev}/\text{s}\). In this case the speed of the rotor, magnetically suspended in vacuum, reached \(1\,200\ \text{rev}/\text{s}\) without loss of stability. The results of these experiments make it possible to assert that rotors of both small and large dimensions, suspended in this way, can rotate stably at arbitrarily high speeds—up to bursting speeds. At a speed of \(600\ \text{rev}/\text{s}\) the driving torque was removed. The subsequent observed slowing of the rotation was approximately \(2\cdot 10^{-3}\ \text{rev}/\text{s}^{2}\), which corresponds to a frictional moment of about \(10^{-2}\ \text{dyne}\cdot\text{cm}\). The friction is due to: 1) residual gas in the chamber, 2) currents in the disk caused by components of the external magnetic field, and 3) hysteresis in the magnetic rod caused by the same components. After these factors have been reduced to a minimum, the friction can be further decreased. An attempt was made to rotate a plate suspended according to Holmes by means of alternating voltages applied to quadrants fixed beneath the plate \(^{26}\). Synchronization of such an electrostatic motor was achieved in various ways: one of them consisted in automatic control of the voltages by interrupting a beam of light with the rotating plate itself. The speed was \(3\,200\ \text{rev}/\text{s}\).

In concluding the review, let us dwell on the most recent report by McBain, who in recent years has continued to improve his “air top” \(^{27}\). In its defense McBain gives the following arguments: low cost, ease of thermostating the rotor, and, finally, the fact that starting and stopping take only a few seconds. The disadvantages of the instrument are above all the considerable nonuniformity of the force field (the rotor diameter is 37 mm) and the small height of the liquid column—less than 4 mm. However, pos-

The latest model of the “optical” centrifuge (December 1939), according to McBain, gave good results. It differs from the McBain–O’Sullivan rotor by better thermal insulation of the cell: a cover with two quartz windows is screwed onto the upper part of the rotor; this eliminates friction of the cells against air and creates an air layer in front of them. The cells are inserted not directly into the rotor, but into an ebonite sleeve \(0.6\) mm thick, and are thus insulated from the rotor body. Between the quartz walls of the cell there is inserted not an ebonite but a platinum-iridium plate, which improves temperature equalization in the liquid. Finally, the periscope is replaced by metal mirrors fastened on the rotor itself in the cavity of its cone at an angle of \(45^\circ\) to the axis; the bottom of the conical part of the rotor is closed by a quartz window. At a speed of \(1900\) rev/sec and an acceleration of \(200000\,g\), measurements of the molecular weight of hemoglobin gave a result coinciding with Svedberg’s data. Another type of improved rotor (with insulated cells, but with a periscope) gave, for the sedimentation constant of egg albumin, the mean value \(3.64 \cdot 10^{-13}\), whereas the 1935 rotor gave \(2.98 \cdot 10^{-13}\). (Svedberg’s value is \(3.55 \cdot 10^{-13}\).)

In the same work McBain reports a new, simpler type of rotor with photographic recording. This is a solid metal cone with an ebonite attachment, in which there is a radial slit at the top. In the slit is mounted a simple capillary tube containing the liquid under investigation; during rotation the tube is photographed on film without the use of a special camera, shutters, etc. This device gave good values of the constant \(s\) for such proteins as hemoglobin.

LITERATURE

General articles and reviews

I. T. Svedberg, Ind. Eng. Chem. Analytic. Ed., 10, 113, 1938.
II. T. Svedberg, Naturwiss., 22, 1934.
III. T. Svedberg, Ber. d. Deutsch. Chem. Gesellsch., 67, 117 (A), 1934.
IV. T. Svedberg, Koll. Z., 67, 1, 1934.
V. J. W. Beams, Rev. Mod. Phys., 10, 245, 1938.
VI. J. W. Beams, J. Appl. Physics, 8, 795, 1937.

1. T. Svedberg and J. Nichols, J. Am. Chem. Soc., 45, 2910, 1923.
2. T. Svedberg and H. Rinde, J. Am. Chem. Soc., 46, 2677, 1924.
3. T. Svedberg, J. Phys. et Rad., 2, 227, 1931.
4. K. O. Pedersen, Nature, 135, 1935.
5. Henriot et Huguenard, C. R., 180, 1389, 1925.
6. Girard et Chukri, C. R., 196, 327, 1933.
7. Garman, Rev. Sci. Instr., 4, 450, 1929.
8. McBain and O’Sullivan, J. Am. Chem. Soc., 57, 2631, 1935.
9. McBain and Alvarez-Tostado, Nature, 139, 1066, 1937.
10. McBain and F. Leyda, J. Am. Chem. Soc., 60, 2998, 1938.
11. J. Beams, Rev. Sci. Instr., 1, 667, 1930.
12. J. Beams and F. Pickels, Rev. Sci. Instr., 6, 299, 1935.
13. Snoddy and Beams, Science, 85, 273, 1937.

1. Davis, Rev. Sci. Instr., 7, 96, 1936.
2. Shapiro and Butt, Rev. Sci. Instr., 8, 35, 1937.
3. J. Beams, Rev. Sci. Instr., 9, 413, 1938.
4. Beams and Haynes, Phys. Rev., 50, 491, 1936.
5. Beams and Masket, Phys. Rev., 51, 384 (A), 1937.
6. J. Bauer and E. Pickels, J. Exp. Med., 65, 565, 1937.
7. J. Bauer and E. Pickels, J. Exp. Med., 64, 503, 1936.
8. Hoxton and Beams, Phys. Rev., 51, 690 (A), 1937.
9. Beams and Snoddy, Science, 85, 185, 1937.
10. Black, Beams and Snoddy, Phys. Rev., 53, 924 (A), 1938.
11. Holmes, Rev. Sci. Instr., 8, 444, 1937.
12. Holmes and Beams, Nature, 140, 30, 1937.
13. Smith, Phys. Rev., 53, 688 (A), 1938.
14. McBain and Lewis, J. Phys. Chem., 43, 1197, 1939.