HOROWITZ AND LEUPOLD 



217 



15- 



lo— 



5- 



Abscisso Colonies/plate 

 Ordinate Temperature mutants/ 

 lo^secondory colonies 



5oo 



7" 



looo 



"T" 



l5oo 



Fig. 2. Frequency of temperature mutants per hundred secondary colonies as a func- 

 tion of population density. 



the growth factor excretion by the 

 wild type is qualitatively similar to the 

 composition of complete medium. 

 Large amounts of such enrichment 

 would tend to support the growth of 

 temperature mutants which had lost a 

 dispensable function, up to the point 

 of visible colony formation, and would 

 thus remove them from the isolation 

 procedure. On the other hand, smaller 

 quantities of the same enrichment 

 might give these same mutants an ad- 

 vantage in the struggle for survival 

 during the first 48 hours at the mutant 

 temperature and during the second 

 period at the lower temperature when 

 they begin to grow against the heavy- 

 competition of established wild type 

 colonies. The actual situation would 

 be far more complex, in that the sign 

 and magnitude of the selection would 

 differ from mutant to mutant. 



In spite of the indisputable occur- 

 rence of cross-feeding on the plates, 

 however, it can be said that no sys- 

 tematic selection favoring either of the 



two classes of temperature mutants is 

 deducible from our data. In Figure 3 

 is plotted the fraction of temperature 

 mutants which were found to belong 

 to the indispensable class, together 

 with the ranges within which the 

 values would be expected to fall in 95 

 per cent of trials, against the popula- 

 tion density. With the exception of 

 one experiment, the data are consistent 

 with the assumption that these are 

 random samples drawn from a homo- 

 geneous population of temperature 

 mutants in which the frequency of 

 mutants of the indispensable class is 

 between 20 and 25 per cent. In short, 

 with the exception of a single experi- 

 ment in the middle range of popula- 

 tion densities, there is no indication 

 that population density influences the 

 relative frequencies of the two classes 

 of temperature mutants: the various 

 selective forces appear to affect both 

 classes equally. 



We may summarize the E. coli re- 

 sults, then, by saying that the fre- 



