III. CENTRIFUGATION 93 



tion that a Duralumin rotor can be accelerated almost three times as 

 rapidly as a steel one because of its correspondingly lower specific 

 gravity. Also, at full speed it requires proportionately less barricad- 

 ing to furnish adequate protection to operator and equipment in the 

 event of rotor failure. And besides presenting a considerably simpler 

 machining problem, it has the very great advantage of high conduc- 

 tivity and hence low susceptibility to temperature gradients and con- 

 vective disturbances. 



Selection of an optical system generally resolves itself into a con- 

 sideration of the various refractive index methods, since they in gen- 

 eral offer distinct advantages over the absorption method, which 

 requires knowledge, not too simply or precisely acquired, of the rela- 

 tionship at every level in the cell between concentration and the cor- 

 responding amount of photographic blackening in the image of the 

 revolving cell. As for the refractive methods, the Lamm method 

 using a real scale is undoubtedly subject to the fewest optical errors. 

 If an optically projected scale is used, the zonal errors of the pro- 

 jection system must be negligible or else accurately measured and ap- 

 plied as corrections. The cylindrical lens method involves the great- 

 est number of optical parts, each of which must be precisely posi- 

 tioned and oriented, and is hence most susceptible to error if im- 

 properly designed or assembled. It is essential that the lenses be of 

 good optical quality. Apparently most of the disappointments with 

 such systems have been caused by improper assembly or a lack of 

 accuracy in the cylindrical lens. Because of the Hmited demand for 

 precise cylindrical lenses, there are few optical concerns that have 

 the specialized experience and equipment required for grinding them 

 accurately. If light is paralleled through the cell (to minimize paral- 

 lax errors) and if good lenses are used, focal lengths can be relatively 

 short (of the order of 50 cm. for the camera lens) and the total length 

 of the optical system can be made as little as 3 m. With any well 

 designed system of the refractive index type the limiting accuracy is 

 determined primarily by the optical resolution of the system, and for 

 measuring the sedimentation rates of distinct boundaries, a good 

 cylindrical lens system appears to be comparable in accuracy to the 

 scale method, although the latter may afford slightly greater accuracy 

 in the determination of the actual distribution of material within a 

 boundary (25). In general, the very convenient cylindrical lens 

 method can be made quite adequate for sedimentation velocity stud- 

 ies, whereas use of the scale method may be warranted for equilibrium 

 determinations. 



