DEFINITION OF THE GENE 333 



assumed that the units of crossing-over or genes are such particles. 

 Moreover, the genes, hke the chromomeres, are chemical indi- 

 viduals. The genes have characteristic properties of mutation, 

 the chromomeres of attraction. We still do not know, however, 

 what the relation may be between the differences from which the 

 gene is inferred and the particles that are observed. We know that 

 particles having a linear order may change in three very different 

 ways to give hereditary differences. 



First, they change externally by loss or reduplication (inter- 

 genic change) to give those differences in proportion that are known 

 to have a physiological effect. By such a change the " bar " gene 

 difference arises in Drosophila (Sturtevant, 1925 ; Bridges, 1936). 



Secondly, they must be supposed to change internally (intra- 

 genic change) to give those differences that distinguish the different 

 genes within the hereditary complement. Differentiation cannot 

 have arisen merely by quantitative change. 



Thirdly, it has now been shown beyond doubt that differences 

 arise merely by changes in relative position of genes. This position 

 effect must depend on a localisation of the products of gene action. 

 It is responsible for two bar genes in one chromosome having a 

 different effect from one bar gene in each of two chromosomes 

 (Sturtevant, 1925). It is also responsible for changes occurring 

 frequently when chromosomes are broken and their parts recombined 

 in a new way (Muller and Prokofieva, 1934). 



Now we see that when we speak of a gene as a particle whose 

 changes are responsible for these differences we may not always 

 mean the same thing. And this difficulty becomes more serious 

 when we consider that the gene is a unit of crossing-over. As a 

 rule, the unit of crossing-over agrees with the unit of mutation. 

 But this is not always so. The reason is clear. A structural change 

 such as inversion will suppress crossing-over at the same time that 

 it determines a genetic difference by position effect. This is 

 perhaps the explanation of elongatus-vitellinus crossing-over in 

 Lebistes (Winge, 1934). The unit of crossing-over is therefore liable 

 to be increased by the act of mutation. The old criteria of genetic 

 structure therefore require re-examination. 



It is in these circumstances that Muller and his collaborators 



