552 CELL MECHANICS 



mere is therefore an interference centre. Thirdly, he finds that the 

 mean distance of the proximal chiasma from the centromere is a 

 function of the length of the arm. There is therefore a proximal 

 region with no crossing-over which no doubt when long established 

 has become inert. 



The double and contradictory action of the centromere as an 

 interference centre and an interference-inhibitor can be regarded in 

 the light of the present mechanical hypothesis as due to a single pro- 

 perty, viz., that the centromere reduces or even abolishes the torsion 

 on the adjacent chromatids at the moment of crossing-over owing 

 to a lack of longitudinal cohesion and a free rotation of bonds. 

 This property is implied by the failure of spiralisation at the centro- 

 mere. Hence it interferes with crossing-over in its neighbourhood 

 and at the same time prevents interference acting through it, since 

 the strain on either side of it can be reduced no further. The 

 action of the centromere is probably greater than that of a cross- 

 over in reducing strain, since it seems (at least in Drosophila) to 

 interfere with crossing-over for a greater distance than crossing- 

 over itself does. This means that crossing-over leaves an average 

 unreleased strain greater than zero, a conclusion which can be 

 deduced directly on the present hypothesis from the necessary 

 discontinuity in uncoiling, one revolution for a pair of chromatids 

 being the indivisible unit. 



The action of the centromere implies two underlying properties. 

 First, as Mather points out, that crossing-over should begin in its 

 neighbourhood. Such a proximal priority has been observed cytolo- 

 gically in Fritillaria (D., 1935 b) ; it is conceivable that in other-- 

 organisms the ends might enjoy priority, and crossing-over show a 

 similar end-relationship. Secondly, the torsion-strain between the 

 chromosomes and chromatids must be released at the centromere by 

 the absence of longitudinal cohesion, while separation is beginning. 

 Since there is no evidence of any plane of symmetry, or at this stage 

 of any preparation for division, in the centromere, positional speci- 

 ficity must certainly be lacking. 



The special length-frequency relationship found in Stenobothrus 

 (Ch. VII) can be accounted for on the same assumption that the 

 average distance of the first chiasma from the centromere is a 



