188 THE PHYSICS OF VIRUSES 



This indicative experiment has been followed by a tedious but 

 powerful study by Lui'ia (1951). Like other mutations, bacterial 

 virus mutations take place spontaneously. This is related to the 

 rearrangement of a molecular order in consequence of the ther- 

 mal agitation of a large molecule. Whatever the cause of such 

 mutations, they occur, and they do so while the phage is active, 

 namely during the infective time in the host. Now the presence 

 of a mutation can be detected by the ability to produce lysis in a 

 host which is resistant to the parent strain, and so produces 

 plaques. The use of mixed bacterial cultures, B and i?/2, where 

 B/^ is resistant to T-2 but not T-2/i, is a convenient device, for, 

 unless both T-2 and T-2/i are present, a cloudy plaque, contain- 

 ing developing -B/2, is formed. The rapid lysis mutant T-2r can 

 also be recognized by the larger plaques formed. 



Luria's experiment consisted in patiently observing the fre- 

 quency of occurrence of two mutants, T-2r and T-2?r, in 23,000 

 bursts of bacteria infected with a strain T-2L. In each burst, the 

 number of mutants was observed, and the fact that in a fair 

 number of cases 10 or more mutants per burst occurred was re- 

 corded. In these cases, the identity of each mutant was checked 

 by using them in mixed infections and showing that no new re- 

 combinant forms resulted. To see what the experiment can tell 

 us, consider three possible ways in which these mutations can 

 occur. 



The first, and most obvious, is that any phage, independently 

 of reproduction, can mutate at any time with a definite proba- 

 bility. This is a rather odd idea since it supposes that the mutant 

 phage produces no mutant progeny. It could happen if a master 

 template phage were stamping out replicas and some of the rep- 

 licas subsequently underwent a molecular rearrangement which 

 was capable of making different phage when the changed replica 

 had its turn in a new bacterium. If the average of all mutants 

 turned out to be x per burst, then the Poisson formula should 

 hold, and the number of times that a mutants would occur would 

 be determined by P{7i), where 



