198 THE BEHAVIOUR OF POLYPLOIDS 



than in homozygous diploids ; it is more frequently noticed in 

 haploids and hybrids because of the striking change it produces in 

 them. Gametic doubling or failure of reduction does occur more 

 freely in haploids and hybrids, but for the demonstrable reason that 

 meiosis has proved unworkable — and not in anticipation of its doing 

 so (Ch. X). 



The observations of the relative fertility of auto- and allo- 

 tetraploids provide sufficient reason for supposing that polyploid 

 species will be of the hybrid kind where they depend on their sexual 

 fertility for their reproduction. The evidence of the relationship 

 of the constituent gametic sets of polyploids {i.e., of the extent to 

 which they are alio- or auto-polyploid) is provided by a series of 

 observations which, owing to their importance for the theory of 

 chromosome pairing, may be conveniently considered in relation to 

 " differential affinity." 



4. DIFFERENTIAL AFFINITY : AUTOSYNDESIS 

 In the diploid Primula kewensis the nine chromosomes of P. 

 verticillata pair with the nine chromosomes of P. florihunda, their 

 presumed homologues. In the tetraploid, however, each chromo- 

 some pairs with its identical mate, florihunda with florihunda 

 and verticillata with verticillata. Only exceptionally one, two or 

 three quadrivalents are formed through all four homologues 

 associating. The explanation of this difference lies in the method 

 of pairing amongst the similar threads at zygotene and of chiasma 

 formation in the paired threads at diplotene, described in polyploids. 

 If florihunda and verticillata chromosomes can only pair in half 

 their length at pachytene this will probably suffice to ensure regular 

 metaphase pairing by the formation of one chiasma in the diploid 

 hybrid. In the tetraploid their chance of forming a chiasma would 

 be reduced by two-thirds if the similar parts continued to pair at 

 random. But from what we now know of the structural differences 

 between species we cannot assume that the differences in undefined 

 hybrids are so simple as here suggested. Probably such portions 

 of the corresponding chromosomes as can pair are distributed in 

 several segments. There will then be several differences of linear 

 sequence (or changes of homology) in the dissimilar chromosomes. 



