THE MODIFICATIONS OF THE CHROMOSOME CYCLE 7I 



in twos, and these bivalents are separated normally at anaphase, giving 

 diploid balanced gametes which function. The theoretical segregation^ 

 of genes in such a system can be calculated on various assumptions, of 

 which the simplest, appropriate for a strict autotetraploid, is that the 

 similar chromosomes pair at random and segregate as units. An auto- 

 tetraploid duplex for a factor A (i.e. AAaa) will have association into 

 two groups of Aa twice as often as into one A A and one aa. From the 

 first association we should get equal numbers of AA, aa and Aa, Aa 

 pairs of gametes, while the latter would give all Aa, Aa pairs. Thus the 

 total gametic output would be i AA : 4 Aa : 1 aa. The ratios of off- 

 spring actually observed in some plants (e.g. tetraploid Primula sinen- 

 sis) agree fairly well with the predictions on this basis. Another theo- 

 retical possibility, is, however, sometimes realized. It has been shown 

 (p. 106) that it is only in the region of the centromere that the paired 

 homologous chromosomes are regularly separated at the first division, 

 whereas in a region distant enough from the centromere for crossing- 

 over to occur between them, portions of two sister chromatids may 

 become attached to different centromeres and thus eventually get into 

 the same gamete. This gives the possibility of random chromatid 

 segregation, which will occur in a region so far removed from the 

 centromere that it becomes an equal chance which centromere a 

 chromatid remains attached to after crossing-over had taken place. 

 The consequences of this type of segregation have been discussed on 

 p. 105. 



2b. Allopolyploids 



Any increase in the differentiation between the various basic sets in 

 a polyploid decreases the chance that the zygotene pairing will be 

 purely at random and increases the chance that it will be strictly in 

 pairs of exactly similar homologues. If a tetraploid A ABB was ori- 

 ginally derived from a hybrid AB, the pairing of A with A and B with 

 B may be much more frequent than that of A with B, the actual rela- 

 tive frequencies depending on how alike the A and B chromosomes 

 are. Even if an A chromosome pairs to some extent with a B in zygo- 

 tene the length through which it is paired with its A homologue may 

 be so much greater that the chiasmata nearly always associate together 

 the two ^'s and the two B's at metaphase. In such a case, we can speak 

 of a differential affinity of the A chromosomes and the B chromosomes. 



Metaphase association of the chromosomes of the two exactly 

 ^ Haldane 1930a, Mather 1933, 1936b. 



