i84 THE BEHAVIOUR OF POLYPLOIDS 



but it is also heterozygous inasmuch as its complement contains 

 two dissimilar pairs of gametic sets. How this affects its behaviour 

 we shall see later. But first we must make an arbitrary but con- 

 venient distinction between different kinds of allopolyploids. On 

 the other hand, there are those derived from doubling in defined 

 structural hybrids ; where these are interchange heterozygotes 

 pairing occurs between parts of structurally dissimilar chromosomes, 

 and associations are formed correspondingly larger than those of 

 the diploid (Ch. V). This group has been considered earlier because 

 it agrees with autopolyploids in having no differentiation between 

 the corresponding chromosomes such as seriously interferes with 

 their pairing at meiosis. 



On the other hand, the typical allopolyploids are derived from 

 undefined hybrids ; they show a gradation in behaviour from those 

 with slight to those with strong differentiation between the 

 homologues derived from opposite parents. At the extreme in one 

 direction is such a hybrid as Crepis rubra x C. fcetida in which the 

 difference between the parents is so slight that the diploid has 

 complete pairing and the tetraploid derived from it behaves like an 

 autopolyploid and occasionally forms five quadrivalents. In an 

 intermediate position is the hybrid Primula kewensis : in the 

 diploid pairing is complete, but in the tetraploid derived from it 

 only a small proportion of quadrivalents are formed. At the 

 extreme in the other direction is Raphanus-Brassica. In the 

 diploid no pairing occurs ; in the tetraploid there are no 

 quadrivalents ; pairing is simple, complete and regular. Both the 

 Primula and Raphanus-Brassica types breed approximately true 

 and are highly fertile. They are, therefore, with a slight qualification 

 functionally diploid. Frequently from crosses between such 

 allotetraploids, through doubling again, oct opioids have been 

 produced, and from crosses with diploids the triploids resulting have 

 given rise to hexaploids. These polyploids must be considered 

 along with the original tetraploids both in regard to origin and 

 behaviour. The}^ may be, in relation to their immediate allopoly- 

 ploid parents, themselves autotetraploid or allotetraploid. Thus the 

 allohexaploid species, Solanum nigrum (2n = 72), yields an auto- 

 tetraploid sport by doubling {2n = 144), while its pentaploid 



