SIZE, SHAPE, AND HYDRATION OF VIRUSES 27 



If nonsplierical viruses are concerned, then the frictional 

 drag coefficient is no longer Qirrja but can generally be written 

 as /, so that the more general equation for the diffusion constant is 



D = kT/f (2.8) 



Diffusion constants, in practice, are measured by two methods. 

 In the first, a sharp boundary is established by using a double 

 cell, the upper part of which can be slid across until it forms an 

 accurate continuation of the lower part. The two parts are 

 filled separately, the lower with virus in buffer solution and 

 the upper with buffer solution alone. When proper temperature 

 equilibrium is established (the diffusion must be carried out 

 under thermostatic conditions), the two parts of the cell are 

 brought into line. The rate of diffusion can be measured by a 

 refraction technique due to Lamm and Poison (1936) which 

 consists of photographing a scale, by light from a mercury arc, 

 through the cell. The boundary is marked by a change in refrac- 

 tive index, and this causes a shift in the apparent position of the 

 scale. From the amount of shift at different positions in the 

 cell, the distribution of virus can be estimated. Initially this is, 

 of course, a step function, but, as diffusion progresses, the dis- 

 tribution becomes that of an error integral, and by evaluating 

 the constant of this function as time passes, the diffusion con- 

 stant can be obtained. Details are given by Neurath (1942). This 

 method requires a rather elaborate, carefully made cell with a 

 good optical system and a good thermostat. It also necessitates 

 homogeneous, pure virus. Four determinations of diffusion 

 constants have been made by this method: bushy stunt virus 

 by Neurath and Cooper (1940), tobacco mosaic virus by Lauffer 

 (1944), southern bean mosaic virus by Miller and Price (1946), 

 and rabbit papilloma virus by Neurath, Cooper, Sharp, Taylor, 

 Beard, and Beard (1941). Of these, there is excellent reason to 

 believe that bushy stunt and southern bean mosaic viruses are 

 accurately spherical, rabbit papilloma is probably spherical, and 

 tobacco mosaic virus is definitely not spherical. The values 

 obtained for the three spherical viruses, together with the derived 

 radii, are given in Table 2.2. 



