THE MUTATION THEORY 497 



are really due to crossing over. The work of Frost on stocks has shown 

 that a precisely analogous situation exists in that form also, and G. H. 

 Shull is obtaining direct evidence for the same conclusion in the eve- 

 ning primrose itself. In any event, it must be granted that so long as 

 this interpretation cannot be definitely refuted, these variations can- 

 not be used as examples on which to base our theory of gene change. 

 In place, then, of the elaborate system of conclusions which has derived 

 its support chiefly from the results in the evening primrose, it will be 

 necessary for our present theory of gene change to erect an independent 

 structure, built upon an entirely new basis. 



The data upon which the new theory must be built consist of two 

 main sorts, which may be called direct and indirect, (i) In the cases 

 giving the direct evidence, the occurrence of the gene change can be 

 proved, and it is possible to exclude definitely all alternative explana- 

 tions, such as contamination of the material, emergence of previously 

 "latent" factors, non-disjunction, etc. So far, the only considerable 

 body of such evidence is that gotten in the Drosophila work, where 

 mutations have (in this sense) been actually observed in at least loo 

 loci. Considered collectively, however, there exist in other organisms 

 enough scattered data to afford ample corroborative evidence for the 

 generality of occurrence of mutations like those observed in the Dro- 

 sophila work. In addition several specially mutable genes have been 

 found in a number of plants (as well as in Drosophila) that are giving 

 highly valuable information along their particular lines. And a num- 

 ber of selection experiments that have been performed on non-segregat- 

 ing lines of various organisms have also given us direct evidence, if 

 not of the frequency, then at least of the infrequency, of mutations. 

 (2) As for the indirect data, these may be gotten by examination of 

 Mendelian factor-differences of all kinds, on the assumption that they 

 must have arisen through mutation. Although this assumption can be 

 shown to be fully justified, these cases cannot provide information 

 concerning the manner of origin of the mutants, nor can they furnish 

 a reliable index of the frequency of mutations, since the mutant genes 

 may have been subjected to an unknown amount of selective elimina- 

 tion or selective propagation before the observations were taken. As 

 for the still more indirect data, derived from studies of phylogenetic 

 series and comparisons between different species, genera, etc., these 

 occasionally give suggestive results, but where crosses cannot be made 

 or where the differences cannot be traced down to the individual genes, 

 such facts can seldom lead to trustworthy genetic conclusions. 



On these various data, duly weighted, we may found our new muta- 



