THE MUTABLE UNIT OF HEREDITY 



51 



propertA of each mutable unit; these units are the indixidual members 

 of the genot^pe. 



Adaptive randomness 



In most of the exp)erLments so far described, the mutants have been 

 selected on a special medium. For example, str-r mutants are identifia- 

 ble onlv in the presence of streptomycin. But we could supi>ose that 

 in the course of testing for mutants the streptomycin itself induced the 

 change. How then do we know that the str-r mutants arose during 

 previous growth in the absence of streptomycin!^ 



The fluctuation analvsis and the distribution of numbers of mutants 

 in different cultures answer this question for us. Were there a small 

 chance that each cell exposed to streptomycin would mutate, the distri- 

 bution of numbers of mutants in different cultures should follow the 

 Poisson formula. It is a property of this distribution that the variance 

 ( (T~ ^ equals the average. \\ e ha\ e seen that the variance of obser\ ed 

 numbers of mutants is manv times the mean Table 2.2 *; the distribution 

 is not Poisson. Therefore we must reject the hypothesis that the muta- 

 tions were induced bv streptomvcin: instead thev occurred during the 

 growth in the absence of streptomycin which preceded exposure. 



200- 



100- 



Coefficient of 

 correlation (r) 



-0.055 



100 2d: 



Number of Methioninetess Bacteria (^ij:"*" 7Tu-t~) 



FIGURE 2.6. Comparison of the number of single mutations of two sorts occurring 

 in a double auxotropfi of E. coli that is both histidinetess and methionineiess \^bis 

 met Ryon unpublisf>ed . 



