CH. X] SUMMARY 147 



Twort found that B. typhosus grown in a lactose medium 

 retained its character as a non-fermenter of lactose for two 

 years before variation occurred. Eyre and Washbourn found 

 that to raise a particular strain of an avirulent saprophytic 

 pneumococcus to full virulence by animal passage, no less than 

 fifty-three successive inoculations were required. Characters 

 which persisted for periods of two, five and nine years, and 

 withstood a series of over fifty passages through an animal 

 body, might well have been regarded as "permanent." They 

 were, however, only "apparently" so. 



Certain principles which govern the stability of impressed 

 variations can, however, be discerned. 



1. The variation may affect all the members of a strain 

 or only certain of them. In the latter case an apparent 

 reversion is obviously more likely to occur. The rapidity with 

 which this apparent reversion takes place will depend upon 

 the comparative rate of growth of the unaltered organisms 

 and the variants. If the new character is of advantage to the 

 organism it will enable the variants to multiply more quickly 

 and they will gradually get the upper hand. Apparent re- 

 version will not take place as long as the new character 

 continues to confer an advantage upon its possessors but when 

 this ceases to be the case the organisms possessing the new 

 character may disappear and reversion to the original type 

 appear to take place. For example, the acquirement of viru- 

 lence by some members of a non-virulent or feebly-virulent 

 strain, when this is injected into the living body, gives these 

 variants an advantage as long as they are in the body. If the 

 mixed strain is grown on artificial media the advantage is 

 done away with and the unaltered bacteria, other things being 

 equal, have now as good a chance as the variants of increasing 

 their numbers and the variants may disappear. 



We have used the expression " apparent reversion " for it 

 is evident that, unless every member of a strain acquires the 

 new character, the loss of that new character by the strain may 

 be brought about by the dying out of the variants without a 

 single organism having actually "reverted." This fallacy can 

 be readily excluded if care be taken at each step to ensure that 



10—2 



