144 STRUCTURAL HYBRIDS 



associated at both ends {e.g., Tradescantta, D., 1929 c ; and Datura, 

 Blakeslee, 193 1). This is due to the fragment usually, on account 

 of its shorter length, forming only one chiasma at prophase. 



(c) Where the fragment corresponds to an intercalary part of a 

 major chromosome, it forms a lateral chiasma, after terminalisation, 

 with the part of the major chromosome with which it is homologous 

 {Tradescantta, Fritillaria, Aucttba, Lilium, cf. Figs. 38, 39, 74 and 80, 

 and Plate IV). 



{d) Fragments which are a small fraction of the length of the large 

 chromosomes fail to pair in a proportion of cases. This failure is 

 expected in organisms with a chiasma frequency proportional to the 

 lengths of the chromosomes, since the frequency of the chiasmata 

 will be less in the short new chromosomes than in the old larger 

 ones, and they will therefore fail to form chiasmata and associate 

 at metaphase, although doubtless associated earlier like the unpaired 

 third chromosomes of triploids. 



The frequency of their pairing will be predictable within limits, 

 on this assumption, when three conditions are determined : (i) the 

 chiasma frequency of the major chromosomes ; (ii) the relative 

 length of the fragments ; and (iii) the " condition " of the fragments, 

 i.e., whether the fragments correspond to parts of the major 

 chromosomes and whether they are in themselves disomic, trisomic, 

 and so on. 



The limits to exact prediction are set, first, by our not knowing 

 whether pachytene association is hindered, on the one hand, by the 

 difficulty of exceptionally small chromosomes finding one another 

 during pairing at zygotene and consequently being interrupted by 

 the time limit, or facilitated, on the other hand, by their freer move- 

 ment. In organisms with a wide size range in their normal comple- 

 ment and non-localised chiasmata, the evidence favours the second 

 view, for although pairing is regular for long and short chromosomes 

 alike, the shorter chromosomes usually have, as we saw earlier, a 

 higher chiasma-frequency relative to their length. Secondly, 

 limits to exactness are set by our not knowing the frequency with 

 which two chiasmata will be formed in any given fragment. A 

 high variance is taken to mean a low interference {q.v.) and will 

 reduce the frequency of association {cf. D., 1933, on Secale). 



