MANNER OF PRODUCTION OF MUTATIONS 477 



derived from the resulting values, extrapolation of this straight line left- 

 ward from the lowest value is found to go, as nearly as would be expected 

 in view of the statistical error, through the point representing zero muta- 

 tion frequency at zero dose. This means that the frequency of the 

 induced mutations is (within the limits of statistical accuracy of the 

 work) directly and simply proportionate to the dose. Thus the evidence 

 shows that the induced-mutation-rate-dosage curve is either a continuous 

 straight line starting at the base level, zero, or at least a straight-sloped 

 "flight of steps," individually too small to be detectable, which start at 

 zero, and that in any case there is reason to conclude that its shape and 

 direction are no different at the lowest doses than at moderate ones. 

 The only reasonable interpretation of this result is that each point muta- 

 tion is the result of a single ionization or activation or of a single natural 

 group of them, comprising a number no larger than that present at the 

 lowest doses used, and that these individual hits (whether they be single 

 changes or clusters), in their turn, neither cooperate nor interfere with 

 each other in the production of mutations. 



It is true that, in certain experiments on the frequency of induced 

 visible mutations at two selected loci in Drosophila, Bonnier and Liining 

 (1949; see also Bonnier et at., 1949) have obtained data which, taken at 

 face value, would indicate a deviation of an opposite character to that 

 postulated by the critics of a linear relation, namely, a straight line point- 

 ing to a frequency at zero dose above that of the control. This would be 

 the case if, besides those induced mutations whose frequency is propor- 

 tional to dose, there was also a small residuum of them which maintained 

 a constant frequency at all doses used but which disappeared when no 

 radiation except that of the natural background was present. These 

 authors suggest that this apparent residuum may have been composed of 

 mutations which were induced by the radiation in a different manner 

 from the others, e.g., by an indirect effect through the cytoplasm, and 

 that for these there was a saturation effect reached even by the lowest dose 

 used. Perhaps the radiation had produced a mutagenic substance from 

 a substrate present in very limited quantity, or the mutagen may (like 

 hydrogen peroxide) have been destroyed when produced above a certain 

 low concentration. However, these experiments might be readily 

 explained in a number of other ways without going so far afield, as by a 

 negative variation in the frequency in the controls (which were carried 

 through at a different time than the treated lots) or, more likely, by the 

 existence of a stronger selection in the control than in the treated lots 

 against the survival of the mutant larvae and pupae (which, being 

 homozygous for "visible" genes, must have had distinctly reduced 

 viabilities), caused by the much greater degree of crowding which seems 

 to have existed in the control cultures. 



The inference, based on the dosage studies, that a gene mutation is 



