530 CELL MECHANICS 



most frequent in many trivalents and quadrivalents. Now, if we 

 consider the two centromeres of a bivalent as lying in a more or 

 less predetermined axis, the three centromeres of a convergent, 

 and therefore biaxial, trivalent must be regarded as having modified 

 one or both of its axes. In other words, the centromeres modify 

 the action of the spindle, and since this action depends on structure, 

 they must be supposed to modify its structure. The same modifica- 

 tion is sometimes seen with interlocked bivalents. But this modi- 

 fication is limited to a twist of io° or 15°, as is shown by the dis- 

 cordant type of arrangement. Here the orientation of two members 

 of a quadrivalent succeeds at the expense of the other two, which 

 owing to the rigidity of the interstitial chiasmata in the configuration 

 are prevented from entering into any axial relationship. The same 

 rigidity is probably responsible for the indifferent type, where an 

 odd member of a trivalent lies indifferently with regard to its 

 partners. 



The effect of the centromeres on the spindle may also be shown by 

 what happens in their absence. Where spindles are formed without 

 any chromosomes they are narrower than the normal (Fankhauser, 

 1934, a and h). Where all the chromosomes are univalent and 

 orientation and congression fail, the spindle gradually stretches 

 and becomes longer than it would be at any stage in normal develop- 

 ment. The cumulative effect of the polar repulsions in orientating 

 the spindle, which itself increases the effectiveness of the repulsions, 

 lengthens the spindle so that it is bent round on itself in a confined 

 cell (Matsumoto, 1933, and Katayama, 1935, on Triticum ; Dob- 

 zhansky, 1934 on Drosophila ; Morinaga and Fukushima, 1935, on 

 Oryza ; Brieger, 1934, on Nicotiana ; Bergner ei al., 1934, on 

 Datura). 



The efect of the chromosomes, therefore, is to broaden the 

 spindle when they come on to the metaphase plate. This effect may 

 be inferred in another way, from the behaviour of univalents in 

 Triticum a.nd ^sculus (KihsLia., ig2g ; Upcott, 1936 &). They come 

 on to the plate after the bivalents and form a ring outside them. 

 If they had not widened the plate they could not have found room 

 on it. While the spindle is therefore entirely independent of the 

 chromosomes in its origin it depends on a centric reaction for the 



