Methods are described whereby the frequency of mutations induced in Drosophila spermatozoa by ultra-violet light may be brought to approximately 5%. Among the means used to increase the mutation rate was the selection of those individuals in which, as shown by their earlier death, the radiation had penetrated more effectively. It has been shown by the use of a combination of filters that the region in which the mutation-inducing effect of ultra-violet light begins lies between 320 and 300 m$\mu $, as expected for nucleic acid absorption. Further results, with an improved genetic technique whereby the production of gene mutations and gross rearrangements could be studied in the same individual, have confirmed our earlier conclusion that, as compared with X-rays of the same gene-mutational strength, ultra-violet rays are ineffective in causing gross structural changes of chromosomes. Preliminary evidence was obtained that ultra-violet rays are similarly ineffective in the production of minute structural changes, and that (in Drosophila) they also fail to produce, by 'simple breakage', chromosome fragments that are capable of surviving ('terminal deficiencies'). It may be concluded that activation of the nucleic acid by ultra-violet is ineffective in producing breakage of the chromosome, although it is effective in producing gene mutations. The gene mutations produced by ultra-violet must occur secondarily, as a consequence of transfer of energy from the nucleic acid, resulting in turn in a change of the distinctive gene material, which is probably of protein nature. It is therefore illegitimate to calculate the size or number of genes from the frequency with which gene mutations are produced by a given dose of radiation. The gene-mutational change apparently differs in some essential way from the change involved in demonstrable minute rearrangements, but the minute rearrangements appear to be essentially similar to the gross rearrangements, involving a similar kind of thoroughgoing breakage of the chromonema. The effect of ultra-violet in discriminating between gene mutations on the one hand and both minute and gross rearrangements on the other hand, lends some support to the conception that the 'genes' represent definite segments of the chromonema, and that the connexions between the parts of these segments are different in kind from the connexions between the segments themselves. A possible means of attack on the mutation theory of cancer, derived from the above conclusions regarding the mechanism of action of ultra-violet light, is pointed out.