96 E. G. P I C K E L S 



of spreading due to diffusion alone can also be ascertained, at least 

 with a vacuum ultracentrifuge, by decelerating to a very low speed 

 before the boundary reaches the bottom of the cell and allowing it to 

 diffuse, with negligible sedimentation, at this low speed (see Fig. 7). 

 With large particles, such as viruses especially, aggregation appears 

 to be a common phenomenon {54) and may produce either well de- 

 fined multiple boundaries or a single blurred boundary with the 

 trailing edge having a rate characteristic of the primary particles. 

 For the first photograph {A) in Figure 7, the speed of the ultracen- 

 trifuge was reduced to a very low value, after the required boundary 

 displacement had been accomplished, in order to minimize boundary 

 sharpening that occurs during rapid sedimentation and to permit 

 Iree diffusion of the material for a determination of the diffusion con- 

 stant. Displacement of base line toward the left in the second picture 

 {B) is due principally to hydrostatic compression of the fluid by the 

 centrifugal force. 



A boundary may be made more or less sharp and its rate affected 

 if the primary particles are in equihbrium \vith dissociation or as- 

 sociation products whose relative concentration depends on the con- 

 centration of the primary particles {1, p. 28) Many proteins appear 

 to suffer partial dissociation in concentrations much below 0.5% and 

 some association in concentrations much above 1%. This limits the 

 range that can be employed usefully for extrapolating the results to 

 zero concentration. 



3. Accuracy and Limitations of the Method 



With an ideal material (high molecular weight and stability) de- 

 terminations of sedimentation rate can be repeated with an average 

 deviation of the order of 1% or better when concentrations of at 

 least a few tenths of one per cent are used and the optimum condi- 

 tions realizable in actual practice prevail. In most cases the greatest 

 error probably occurs in the measurement of cell temperature. At- 

 tainable accuracy necessarily decreases as more diffusible or less 

 stable material is investigated. In a typical study of the smaller 

 serum proteins, for example, an average deviation of several per cent 

 is not uncommon and an extreme difference between individual de- 

 terminations of as much as 10% is not unexpected. If a sedimenta- 

 tion constant is to be determined with any degree of precision, at 

 least several different determinations must be made. Even when a 

 mean value is based on many careful determinations made with a 



