Discussion and Summary Studies have been made on the ultracentrifugal fractionation of chorio-allantoic fluid from embryos infected with the PR8 strain of influenza A virus. By means of alternate low- and high-speed ultracentrifugal cycles, a particulate component of 77.6 mμ diameter and a sedimentation-constant of S20° = 724 × 10-13 has been isolated. An identical component has been obtained by ultracentrifugation of eluate of red blood cells agglutinated by virus-infected chorioallantoic fluid and of virus adsorbed on calcium phosphate. Tests of the various fractions made quantitatively within the limits of the methods by red blood cell agglutination, specific precipitation, complement fixation and infectivity have indicated an association of the virus with these particles. Two papers describing experiments with influenza virus, previously reported briefly (2, 27), were recently published, one by Chambers and Henle (7) and the second by Chambers, Henle, Lauffer and Anderson (8). These authors reported the concentration of a material of 10 to 12.5 mμ particle-diameter which was considered to be the virus of influenza. In addition, a component with a sedimentation-constant of S20° = ca800 ± 100 × 10-13 was observed in sedimentation-velocity studies, which “was shown to consist principally of aggregated particles of the more disperse fraction.” In general, the methods employed in this laboratory were somewhat similar to those used by Chambers and his associates, yet the conclusions reached were in some respect wholly dissimilar. Although a complete explanation of the differences in conclusions is not apparent, certain aspects of the present work will be discussed with consideration of the results of Chambers and his co-workers. The chorio-allantoic fluid used for the present purification-process was not frozen, since freezing might have resulted in loss during initial clarifying low-speed centrifugations. It has been shown that a precipitate is developed in chorio-allantoic fluid on thawing from the frozen state, and this has been made the basis for concentration (4) of the virus. After preliminary experiments, the fluid was dialyzed immediately on removal from the embryos, since this procedure eliminated the urates which otherwise sedimented in the first ultracentrifugal cycle. The use of calcium chloride was based on previous studies on equine encephalomyelitis virus (19), wherein it was found difficult to demonstrate in the electron microscope virus particles in the numbers expected on the basis of known concentration in the preparations. While studying the influence of various salts on the Eastern strain virus, preparations were made by dilution with calcium chloride solution. Films of such preparations showed a difference from those diluted with water or saline in the even dispersion of the approximate expected number of particles in electron micrographs. The contrast of the particle-images was greater and salt crystals were infrequent. It was thought that the phenomenon might be related to the hygroscopic properties of the salt in preventing violent drying currents and consequent removal of virus from the field with salt crystallization. Another possible explanation for this effect is the known (28) dispersing action of calcium chloride, which may have acted to maintain uniform suspension and spreading of the virus on the film. The same phenomenon was observed with the concentrates associated with influenza virus, for, while the large particles were routinely demonstrable in the presence of calcium chloride of 0.023 M concentration, satisfactory demonstration of this component in water preparations was difficult. The inclusion of calcium chloride in the purification-process was based on the previous observations and on additional findings in the present work. Ringer solution was first used as the solvent, but it was found that when ultracentrifugal concentrates were taken up in Ringer solution alone, a large part of the viral activity was lost in the low-speed cycle in the ultracentrifuge or the angle centrifuge. In the hope of minimizing this loss, calcium chloride in 0.023 M additional concentration was included in the solvent for the entire procedure. Under these conditions, the viral activity lost in low-speed centrifugation was markedly reduced, indicating more complete re-dispersion of the particles packed by ultracentrifugation at 27,000 g. Attempts were made to purify the concentrate, using 0.1 M salt solution. At this concentration of salt, the particles were partially insoluble and nearly all of the viral activity was lost in the low-speed runs. Control studies showed that calcium chloride, as used, did not affect the specific relations in the biological tests employed. The experiment combining elution from red blood cells and ultracentrifugal concentration was carried out on the basis of possible selective adsorption (15), in the hope of obtaining thereby the specific viral particles freed from extraneous material present in chorio-allantoic fluid. The material removed by the agglutinated red blood cells, and subsequently concentrated by ultracentrifugation, was predominantly of about 80 mμ particle-diameter, as shown by the electron microscope (figs. 3 and 4). Small particles were plentiful in the supernate of the chorio-allantoic fluid after red blood cell adsorption, but no evidence of them was seen in the sedimentation-velocity diagrams of the concentrated eluate (fig. 6) and relatively few were seen in the electron micrographs (figs. 3 and 4). The small particles were obtained in profusion from the adsorbed supernatant fluid after subjecting it to higher-speed ultracentrifugation (67,000 g for 2 hours). Little activity was associated with such material, and that seen could be accounted for on the basis of a small amount of the larger component seen in the electron micrograph of figure 8. In addition, ultracentrifuged concentrates of virus eluted from calcium phosphate contained particles of about 80 mμ as the predominant material, as shown by the electron microscope (figure 10), and no boundary corresponding to particles of about 15 mμ was seen in the sedimentation diagram. It is of interest to note that small particles of about 15 mμ have been found in large numbers in association with equine encephalomyelitis virus obtained from the chorio-allantoic fluid of infected embryos. This is illustrated in figure 11, which is the micrograph of a concentrate obtained by ultracentrifugation. There are present the characteristic, relatively large circular virus particles (47.4 mμ) of definite internal structure, in association with the smaller particles. The influenza viral activity was rapidly removed from the dialyzed chorioallantoic fluid at 40,000 g, as is demonstrated in figure 7. About 50 per cent was removed after 10 minutes and about 90 per cent after 30 minutes. The slow approach of the curve to zero could mean that a small amount of the virus did not sediment, or it could mean that this amount was washed from each pellet when the supernates were poured off. The 95 per cent or more which sedimented did so at a rate two to three times that of papilloma virus (26), suggesting a sedimentation-constant two to three times S20° = 278 × 10-13. Examination by the sedimentation-velocity method of the viral concentrates, those obtained from the red cell eluate, those eluted from calcium phosphate and those directly from chorio-allantoic fluids shows the presence in each of a single, slightly diffuse boundary with sedimentation-constant of S20° = 724 × 10-13. This sedimentation-constant is seen to be 2.6 times that of papilloma virus. The results of the present work are in accord with the findings of previous workers in relation to the size of the particle with which activity is associated. From ultracentrifugal data, Elford and Andrewes (22) estimated the diameter of the viral particles to be 87 to 99 mμ. Data obtained in ultrafiltration-studies indicated a particle-diameter of 80 to 120 mμ (24). Hirst (3) found that a large part of the viral activity was sedimented by spinning the chorio-allantoic fluid at 11,500 rpm for 45 minutes. Chambers and his associates (8) recorded that infectivity was associated in part with the component of S20° = 800 ± 100 × 10-13 in the preparation in which the latter was present. From the work of Elford and Andrewes (22), the density of the virus is 1.20. This, together with the sedimentation-constant, S20° = 724 × 10-13, makes possible a calculation of the diameter of the particles, if they are spheres. This value is about 80 mμ, which corresponds closely to the findings with the electron microscope and those obtained in comparing rates of sedimentation of the influenza and papilloma viruses. The possibility of small viral particles being adsorbed on larger particles, as suggested by Chambers et al., obviously cannot be ruled out at present. It would appear from their experiments that this association, if it occurs, was irregular in concentrates obtained from chorio-allantoic fluid. Presumptive evidence was advanced indicating aggregation of the smaller units into forms with S20° = 800 ± 100 × 10-13, but no electron micrograph of such aggregates was shown. In the present work, the particles of 77.6 mμ seen in the electron microscope clearly were not aggregates, but were characteristically formed elements. In micrographs of fully active material, such as those of figures 2 and 4, no particles of 10 to 15 mμ were seen attached to the larger particles. Such would easily have been seen at the magnification of these figures. No definite evidence has yet been produced to show that particles of relatively large size act as specific “carriers” for viral particles of relatively small size.