DELAYED EXPRESSION OF MUTATIONS 469 



With the exception of the mutations to B/3, 4 and B/3, 4, 7, estimates from 

 method 5 are of the order of six to seven times those from method 1, and the 

 lower ratios in the case of these exceptions are in all probability due to the 

 high proportion of slow-growing B/3, 4, and B/3, 4, 7 mutants — Demerec 

 and Fano having shown in their experiments on competitive growth that these 

 mutants grow much more slowly than the parent strain B. 



A crude average of the ratio rate (5)/rate (1) from mutations other than 

 those to B/3, 4, and B/3, 4, 7 shows that estimates obtained by method 5 



Table 10 



Mean mutation rates of strains B and B/r to phage resistance, estimated by methods 1, 2, and 5; 

 data (7/Luria and Delbruck, Demerec and Fano, and the present investigation {table 9), all from 

 series of similar liquid cultures started with small inocula. 



* Note: One experiment containing an exceptional culture with 20,000 resistant bacteria has 

 been omitted from these averages. 



are approximately 6.4 times those obtained by method 1. The significance of 

 this ratio in relation to the extent and variability of the delay between muta- 

 tion and phenotypic expression will be considered in the discussion. 



THE INFLUENCE OF DIVISION MORTALITY UPON ESTIMATES 

 OF MUTATION RATE 



All the methods of estimating mutation rate are subject to a small error 

 because some of the bacteria resulting from a division fail to divide again 

 (Wilson 1922). The number of mutations occurring during a division is thus 

 underestimated, and the number of divisions, where calculated from numbers 

 of viable bacteria, is also underestimated. 



It might be assumed that the apparent variations in the mutation rate 

 during the early part of the growth cycle are due to corresponding changes in 

 the mortality associated with division. That this is not the case will be demon- 

 strated; and corrections to be applied to mutation rates which have been based 

 upon the assumption of unity survival will be considered. 



Where growth has been estimated from the numbers of viable bacteria, a 

 correction may be obtained in the following manner. Let S be the chance of 



71 



