RADIATION AND THE STUDY OF MUTATION IN ANIMALS 1223 



Their data are given in Table 7. It is possible to compute, as they have 

 done, from the frequency of viable, fertile homozygotes in the trans- 

 location types 1-2, 2-4 and 1-3, 3-4, the expected frequency of the 2-3 type. 

 This can be done because there are no lethals in the 1-4 type, and it 

 may hence be inferred that the chance of picking up a lethal effect in 

 chromosome IV is negligible. In chromosome I no lethals are recovered 

 in their experiments, therefore any lethal effects must be due to chromo- 

 somes II or III. The computed and observed percentages for the 2-3 

 type are in close accord. From this, and from the rough determinations 

 of the locus of the point of break available, they conclude that there is 

 "no positive correlation between the region of breakage and the lethals 

 induced at the time of breakage." This depends for its validity chiefly 

 on the determination of locus of break, a determination difficult, if not 

 impossible, to make by genetic methods with the accuracy required. 

 The finer analysis possible as a result of Painter's (128) explorations of 

 the size and differentiation of the chromosome of the Drosophila salivary 

 gland is needed for this purpose. The other point, the agreement between 

 calculated and observed values of viable homozygotes, indicates simply 

 that the occurrence of a lethal effect is a function of the chromosome in 

 which that effect occurs and need not depend on the other chromosome 

 involved in the translocation. The question is a complicated one which 

 is difficult to settle without more detailed data than are at present 

 available, except in a few cases. These concern the study of mutations 

 localized around the breakage points of chromosome rearrangements. 



There are two ways in which such experiments may be conducted. 

 Either the tests are so arranged that a particular mutant effect is detected, 

 and examined for the occurrence of a break; or a particular type of rear- 

 rangement is selected, and then examined for the occurrence of the 

 "mutation." Both permit the determination of whether a given break 

 is associated with a given mutant effect. The latter type of experiment 

 tests in addition the possibility that an effect occurs only in association 

 with a break, but that not all breaks at a locus produce the effect. Four 

 loci in all have been studied in either one or the other of these ways: 

 Bar (X-chromosome, 58.0), brown (II, 104 to 106), bobbed {X, 66) and 

 cubitus interruptus (IV, near spindle attachment). 



At the bar locus, the first data of this kind were supplied by Dob- 

 zhansky's study (31) of a mutant which he called baroid. This repre- 

 sented a mutation from wild type to baroid; it was found to be associated 

 with a translocation whose locus in the X was at Bar, and indeed by all 

 available tests turned out to be a Bar allelomorph. According to 

 Sturtevant (171) no effective normal allelomorph of Bar can be detected. 

 Dobzhansky, making u.se of the demonstrated position effect at the Bar 

 locus, suggested that the appearance of the baroid mutation was an 

 effect of the change of neighbors brought about by the rearrangement of 



