III. CENTRIFUGATION 89 



the latter more nearly represents actuality in most cases and hence it 

 seems unwise, lacking information about the compressibility of the 

 particles, to attempt any correction, especially in view of the fact 

 that the increased viscosity of the compressed medium causes a 

 shght decrease in rate. Neglect of all compressional effects with 

 aqueous solutions could hardly introduce an error of more than 1 or 

 2% under the conditions cited. 



Sedimentation rate is usually determined by photographing sedi- 

 menting boundaries at intervals, reducing each measured incremental 

 displacement to the computed amount per unit field of force (basing 

 centrifugal force on the average radial position during the displace- 

 ment), and averaging all the values obtained. Where multiple 

 boundaries are not resolved until they approach the bottom of the 

 cell, the rate must be based on their effective starting time at the 

 meniscus. An allowance for reflection at the meniscus before clear- 

 ance of the boundaries can be estimated from runs with monodisperse 

 preparations of similar particle size and concentration. The correc- 

 tion is negligibly small when the total sedimentation time is not more 

 than about three hours. When the position of a diffuse boundary is 

 taken as the point of half concentration, a slight error is introduced 

 into the determination of sedimentation rates (as well as into esti- 

 mates of diffusion constants) by reason of the fact that diffusion is 

 taking place in a sector-shaped cell. However this error is very small 

 with the large rotors and cells already described and may be neglected 

 (^p.22). 



3. Determination of Partial Specific Volume 



The equations in Section C were derived on the assumption that 

 V, the partial specific volume of the particle, represents the reciprocal 

 of its density under experimental conditions. However, this is sel- 

 dom known with any exactness because of hydration or solvation of 

 the particle by the suspending medium. This tends to increase both 

 the partial specific volume (unless p > c) and the frictional coefficient. 

 In case of proteins, the amount of hydration is usually of the order of 

 20 to 50%. For particles having molecular weights of about 40,000 

 this may represent an outer shell of water approximately one mole- 

 cule in thickness. In case of some larger particles, such as viruses 

 there appears to be an actual imbibition (53). Attempts have been 

 made to determine the partial specific volume of hydrated particles 

 by centrifuging them in media of different densities and extrapolating 



