RADIATION AND THE STUDY OF MUTATION IN ANIMALS 1219 



As has just been pointed out, the data involve such small numbers of 

 mutations even in very large experiments that it is difficult to use them 

 quantitatively. Especially are they complicated by such possibilities 

 of variation as those which Timof^efT-Ressovsky has demonstrated (page 

 1217) for the white locus. If, in spite of this, the frequency of mutation 

 to a given type is plotted 7^ 

 against the frequency of 

 mutation from it, a sur- 

 prisingly regular inverse func- 

 tion results (Fig. 4). It is 

 difficult to trust such a curve 

 too far; yet taken at its face 

 value it indicates that the 

 frequency with which any 

 gene is obtained as a mutation 

 is a function of its own 

 stability. Thus most of the 

 mutations at the white locus 

 are mutations to white, itself 

 the most stable member of 

 the series. Such a relation 

 might result from a number 

 of simple chemical mech- 

 anisms; it is implied in the 



I. 2 3 4 5 6 7 „ 

 Mutation per Ten Thousand From a Given Allelomorph 



Fig. 4. — The relation between the frequency of 

 mutation to a given allelomorph, and its own 

 mutation rate to other allelomorphs. It should be 

 noted that the probable errors of such low values 

 are in themselves relatively large. Moreover, tho 

 data for the white locus (Table 4), from Timofeeff- 

 Ressovsky (183); for the other loci (Table 5), from 

 Johnston and Winchester (76), are in themselves 

 subject to error which can not be properly evaluated. 

 Well-know^n Galton polygon, Different allelomorphs are not separated, and hetero- 

 SO frequently used as an aeneous data are lumped. Each point on the curve 

 •' represents the frequency of mutation to a given gene 



analogy for mutations in all the available data. A full discussion of the 

 (Morgan, Bridges, and '^''^^^'°''^ involved cannot be given here; but the 

 X, . no T. discovery of the possibility of a general rule for such 



bturtevant, 93; Patterson occurrences seems interesting enough for 



and Muller, 141). But much P^«««"t^ti«"- 



more data are necessary before elaboration of such ideas would be 

 useful. 



THE LOCALIZED OCCURRENCE OF THE MUTATION PROCESS 



-In the foregoing sections it has been made evident that the probability 

 that a mutation will occur is a property of the particular gene concerned. 

 This being so, it becomes of interest to note whether, when two members 

 of an allelomorphic pair are present, both mutate simultaneously, or 

 only one of the two changes. The question at issue may be phrased 

 differently: is the mutation process a local occurrence, or is it the response 

 of a given gene to a general change in environment? If the latter were 

 true, both members of a pair should change at the same time; on the 

 former hypothesis, either one or both might change. 



