IMPLICATIONS OF THE RESULTS WITH ULTRAVIOLET 307 



tion of requiring a bimodal curve for the spectral distribution of effec- 

 tiveness of the interfering action, since radiation in the region of 3100 

 A has very little such effect, at least in bacteriophage reactivation [Dul- 

 becco (14)]. However, it is difficult to explain on other grounds why 

 the falling off in net mutagenic effectiveness at higher doses should be 

 so widespread and so extreme as it has been observed to be. 



But no matter which of the above interpretations, or which combina- 

 tion of them, be adopted, the influence of selective mortality or inter- 

 fering radiation should be negligible at low doses, and at these doses 

 (that is, nearer the orgin of the frequency-dosage curve) the mode -of 

 operation of the positively acting factors should become clear, as is usual 

 for sigmoid curves in general. Moreover, it would become very prob- 

 able that we had reached low enough doses for judging the action of the 

 positive factors by themselves if we found a portion of the curve (to the 

 left of its convex region, and including the lowest doses for which effects 

 were measured) in which, over a considerable range of dosage, involving, 

 for instance, a dosage variation by a factor of 4, the amount of effect 

 (mutation frequency) was found to be proportional to a constant power 

 of the dose. That is, we could assume that at these dosage levels the 

 interfering action had not yet come into play appreciably and that the 

 observed curve of results reflected the influence of only the factor or 

 factors which induced the mutations, in the virtual absence of the coun- 

 teracting influences. 



Unfortunately, it is extremely difficult to get data of sufficient statisti- 

 cal reliability for such low doses. However, if we examine the most 

 detailed published data approaching these conditions, such as those ob- 

 tained by Hollaender and Emmons (25) on Trichophyton, and by Hol- 

 laender, Sansome, Zimmer, and Demerec (27) on Neurospora, we find 

 distinct indications that at low doses the mutation frequency rises faster 

 than a simple linear relation to dose would allow. On the other hand, 

 so far as one can judge, the frequency hardly seems to rise faster than 

 the square of the dose, as it would if two activations were required for 

 a mutation. Yet the data are too meager for a decision either on this 

 question or on that of whether low enough doses to rule out the inter- 

 fering factors have been reached. 



Novick and Szilard (61) have reported measurements of the frequency 

 of three specific types of mutation (to phage resistance) in E. coli on 

 application of varied doses of ultraviolet. In this work the amount of 

 effect caused by "reactivation" by light of longer than mutagenic wave 

 length w^as determined at the same time, for each of the types of mu- 

 tation. It seems unlikely that the viability of mutants of these kinds 

 w^ould have been reduced by the ultraviolet much more than that of 



