UNPAIRED CENTROMERES 413 



nucleus relapses to a resting stage without any movement having 

 taken place. This is a means by which unreduced gametes may 

 arise (Papaver, Ljungdahl, 1922 ; Hieracium, Rosenberg, 1927). 



With very few paired chromosomes and extreme delay in the 

 unpaired chromosomes orientating themselves on the plate, a 

 normally developed spindle increases in length, and, as we saw 

 earlier, is bent round until the poles nearly touch {Triticum, Mat- 

 sumoto, 1933 ; Drosophila, Dobzhansky, 1934 ; Roller, 1934 ; 

 Impatiens, F. Smith, 1935 et alii). 



Evidently the paired chromosomes prevent this lengthening and 

 modify the shape of the spindle, just as convergent associations of 

 three chromosomes modify the direction of repulsion between the 

 centromeres of associated chromosomes in the spindle. 



Reviewing these observations we see that the chromosomes which 

 lag are univalents at the first division and daughter univalents at 

 the second. These chromosomes evidently fail to move, although 

 they resemble the other chromosomes in every respect save in not 

 having partners. Movement, therefore, depends on a reaction— 

 presumably repulsion — of pairs of elements. The presence of the 

 spindle, and of the chromosome in the right position in relation to 

 it, is not sufficient. 



The behaviour of univalents thus shows that the ability to divide 

 at anaphase is due to a change in the organ of division, the centro- 

 mere. This change must be an assumption of bipolarity. It also 

 shows that the time of this change in the centromeres of univalents 

 is different in different organisms in relation to the metaphase, 

 that is, in relation to the development of bivalents. But in all 

 cases the univalents are later than this metaphase in becoming 

 bipolar. The precocity of the metaphase in meiosis agrees well 

 enough with the conditions of separation of bivalents, but it leaves 

 the univalents more or less out of step. The degree of precocity 

 of the metaphase, like that of the prophase, varies from one organism 

 to another. 



(v) The Consequences of Non-pairing. We have seen that chromo- 

 somes fail to pair either on structural or numerical grounds (in 

 haploids, triploids and ordinary hybrids) or on genotypic grounds. 

 Whatever the grounds the same kinds of results ensue, the particular 



