CHIASMATYPE THEORY 253 



between them. We also know that reduction in quahty, i.e., the 

 separation of unHke chromosomes, occurs both at the first and at 

 the second division under conditions which will later be defined. 



While this dispute was still undecided, Janssens (19^9, 1924) put 

 forward the objection that the formation of four spores or gametes 

 at meiosis, the occurrence of two divisions, of pachytene pairing, 

 and of the diplotene looping, would all lack purpose if reduction 

 was achieved at a single division and without crossing-over. These 

 arguments are of great value taken metaphorically ; but Janssens 

 also pointed out that without the assumption of an exchange of 

 segments at the chiasmata the exchange of partner lacked any 

 causal explanation. He pointed out that there was free association 

 of larger numbers of factors in some organisms than there were 

 chromosome pairs ; this was genetic evidence of such exchanges. 

 Applied to Drosophila, Janssens' theory became, as we have just 

 seen, the basis of the successful study of the linear arrangement of 

 genes. 



There was, however, no unequivocal direct evidence either for 

 or against the hypothesis. Two chromosomes went into pachytene 

 association : four chromatids emerged at the diplotene stage. 

 Whether an exchange had occurred no direct evidence could or can 

 reveal. Moreover, at this time the regular structure of bivalent 

 chromosomes, the uniform character of the chiasmata, and the 

 relation of the two was not understood. Janssens therefore sug- 

 gested (1924) several different mechanisms of crossing-over, and 

 Belling, who accepted some of these suggestions (1928 c, 1931 h, 

 1933) left the relationship of the mechanism with the observed 

 structures undefined. 



In recent years three advances in interpretation have removed 

 these difficulties. First, we understand that chiasmata are all of the 

 same original structure (exchanges of partner among four chroma- 

 tids) and differ only through movement to the ends. Secondly, we 

 have evidence of special distribution of chiasmata in polyploids 

 and structural hybrids which enables us to define the genetical 

 change that determines them. Thirdly, we have a large body of 

 comparative evidence from experimental breeding and chromosome 

 observations which enables us to demonstrate the general validity 



