36 BAWDEN AND PIRIE 



the usual methods of isolation, preparations of a specific nucleoprotein 

 can be made with fairly consistent properties, although sedimentation 

 boundaries are usually diffuse and electron microscopy ahvays shows 

 particles with different lengths. More critical methods of fractionation, 

 however, show that the nucleoprotein in the original extracts is hetero- 

 geneous and occurs in particles with ^Addely different properties. If the 

 extracts are subjected to alternate cycles of centrifugation at 30,000 

 and 80,000 g., a series of fractions is separated containing particles 

 with different average lengths and consequently with different physical 

 properties and serological behaviour (Bawden & Pirie, 1945b). The 

 infectivity per unit weight of nucleoprotein in such fractions also is 

 correlated with the mean particle length, the most slowly sedimenting 

 material being almost non-infective and the most rapidly sedimenting 

 being most infective. The particles in all the fractions can aggregate 

 hnearly to give forms that sediment more rapidly and show intense 

 anisotropy of flow. After such changes all the fractions have similar 

 general chemical, phj^sical and serological properties, but they main- 

 tain their original differences of infectivity. Aggregating the small 

 particles does not increase the infectivity; length per se, therefore, is 

 no guarantee of infectivity and it is obvious that particles cannot be 

 assumed to be infective merely because they achieve a certain length. 



The origin and significance of these various kinds of particles is 

 uncertain, but there are various possible interpretations. The simplest 

 is to assume that there is one unique particle capable of initiating in- 

 fection, that this dupHcates itself exactly, and that all variants arise 

 by degenerative changes in this particle, produced in vivo or in vitro. 

 There are many treatments that can destroy infectivity without de- 

 stroying serological specificity, although in vitro the only one knov^Ti to 

 break large particles of tobacco mosaic virus into small serologically 

 active fragments is ultrasonic irradiation (Oster 1947), and under 

 most conditions linear aggregation seems more likely to occur than 

 fragmentation. 



There is no a priori reason to assume that the ability to cause in- 

 fection necessitates a unique structure or that infection must neces- 

 sarily alter protein metaboHsm so that only one specific product is 

 formed. Indeed, the frequent production of clinically distinct mutants 

 shows that all the particles produced during virus multiphcation are 

 not identical. To be infective, particles presumably will need a mini- 

 rtnmi number of specific groups, but some parts of the particles could 

 probably vary without destro;ying infectivity. Indeed this is suggested 

 by the fact that most of the amino groups and some of the phenol and 

 indole groups can be substituted without decreasing infectivity 



