i66 STRUCTURAL HYBRIDS 



advantage to the ends of the chromosomes (which pair first) in 

 pairing, and chiasmata will be formed chiefly near the ends. This 

 is probably responsible for an increase in the frequency of terminal 

 association in Triticum hybrids in proportion to the reduction in 

 frequency of pairing. The terminal association has been described 

 as " loose " pairing in contradistinction to " close " pairing where 

 the pairs have formed several interstitial chiasmata. It is directly 

 determined by the formation of fewer chiasmata, and those nearer 

 the ends than usual (Kihara and Nishiyama, 1930 ; cf. D., 1930 c 

 and 1931 d, and Ch. VI). 



We have seen that univalents, the result of incomplete pachytene 

 pairing, often occur with permissive changes of pairing in a tetra- 

 ploid. This is due to the time-limit in pairing which is responsible 

 for certain parts of long chromosomes failing to get paired before 

 they divide, and for the localisation of pairing characteristic of 

 many species with long chromosomes. A fortiori we should expect 

 that incomplete pachytene pairing would occur in any organism 

 with frequent obligatory changes of partner resulting from a com- 

 plicated structural hybridity. Evidently the reduction of meta- 

 phase pairing in undefined hybrids between species is in part 

 attributable to the action of the time-limit. 



There is another special way in which pairing is interfered with 

 at pachytene in a structural hybrid. This has been made clear 

 by the work of McClintock on Zea Mays. She discovered (1933) 

 in defined interchange, inversion, and deletion hybrids a new 

 principle in chromosome pairing. Chromosomes which have begun 

 to pair at places where they are homologous often continue their 

 pairing along their length irrespective of homology. They pair, it 

 seems, like two pieces of twisted string set free together. The 

 pairing begun by attraction continues by torsion, by the same strain, 

 that is, which produces relational coiling in any case (D., 1935 c, 

 1936 d). This means that if we take the chromosomes of two 

 different species which differ perhaps in numerous small details of 

 arrangement, the greater part of their pairing in the hybrid will be 

 between non-homologous segments. Such pairing does not permit 

 of crossing-over and chiasma-formation so far as we know, and the 

 chromosomes will therefore fall apart unpaired at diplotene. It 



