CROSSING-OVER 551 



lying in one plane on both sides of it. Two chromatids then appear 

 to cross one another at the chiasma, and it has usually been thought 

 that these middle threads were the cross-over chromatids. On 

 the present view the reverse is the case. The middle threads still 

 preserve the relational coiling of the chromosomes which is lost 

 in the outer threads, and which they themselves lose just afterwards. 

 Experiments with wool models confirm the conclusion that the 

 outer chromatids are the cross-overs. They also show that a variable 

 amount of coiling may be lost by each crossing-over, so that an 

 estimate of the proportion lost in this way is not possible. It 

 would appear that the special systems of crossing-over relationships 

 found with and without chromatid interference may depend on the 

 regulation of the relative transverse positions at which breakage 

 may occur (D., 1936 d). 



Various corollaries of this hypothesis follow : (i) Since crossing- 

 over releases, as any movement must do, the strain that determines 

 it, interference is inherent in the assumption of a longitudinal 

 cohesion which will transmit this strain ; (ii) the fact that crossing- 

 over is not reduced in proportion to the length of chromosome in 

 triploids, although the length paired is reduced, is presumably due 

 to the total amount of coiling tension being proportionate to the total 

 length of the chromosomes and not to the length paired ; (iii) the 

 occurrence of interference in crossing-over between chromosomes 

 on the other hand is probably not due to conditions at the time at 

 all but rather to an interference in pairing at zygotene, i.e., to an 

 average capacity for the nucleus to permit pairing before division 

 of the chromosomes, the amount of true pairing (as opposed to 

 post-division torsion pairing) being the variable factor limiting the 

 frequency of crossing-over in individual chromosomes. 



It is now possible to consider the special relationship of the centro- 

 mere with crossing-over in the light of a new analysis of Drosophila. 

 Mather (1936) finds first that there is no evidence of interference 

 between cross-overs on opposite sides of the centromere. The 

 centromere is therefore an interference-inhibitor. Secondly, he finds 

 that the crossing-over in a given arm is specially distributed in rela- 

 tion to the centromere. The distribution is of the same kind as if 

 the centromere were a point of constant crossing-over. The centro- 



