i88 THE BEHAVIOUR OF POLYPLOIDS 



mitosis and by failure of such separation at one of the two meiotic 

 divisions, giving in consequence "unreduced" gametes. Where 

 the tetraploid arises by doubling in the somatic tissue of the first 

 generation (F^) hybrid, the affected shoot of this plant is as fertile 

 as the progeny {Fc^ it gives rise to, or even more so. Where it 

 arises from failure of reduction in both male and female germ- 

 cells in the F-^ their successful union can only be occasional ; the 

 F^ has therefore the characteristic sterility of hybrids and the F^ 

 is the first to regain full fertility (cf. Sansome, 1931). The contrast, 

 it should be noted, is not so striking in the Crepis hybrid because 

 it is not entirely sterile as a diploid ; it gives diploid and triploid 

 as well as tetraploid progeny. Further, it should be noted that 

 frequent syndiploidy at divisions immediately before meiosis will 

 be indistinguishable from failure of reduction except by direct 

 observation. 



Hybrid polyploids that have arisen in experiment may be 

 provisionally classified in accordance with these considerations, and 

 with direct chromosome counts, as in Table 26. 



{Note on Table 26. The female parent is given first. The 

 numbers in brackets are the gametic numbers of the parents. In 

 the first column are the somatic numbers of the offspring before 

 doubling (F^) and in the second column the numbers after doubling 

 (F^ or Fg). Where the failure of reduction occurs in the parent the 

 first column is therefore blank. The initials in the second column 

 represent the sets of chromosomes of parental species. Classes A 

 to D show doubling of the contributions of both parents, classes E 

 to H, of that of only one parent). 



These observations show that allopolyploids have arisen in five 

 ways. In the first, type A, a fertile shoot usually appears which is 

 readily distinguishable in the otherwise sterile hybrid, but the 

 doubling may occur so early in the development that, as in certain 

 cases of haploid parthenogenesis, the whole plant has the double 

 number. In type B, the hybrid is usually highly sterile, but 

 produces, here and there, viable seeds. In some hybrids a few of 

 these seeds have the same chromosome number as their parent, 

 or they result from doubling on one side only so that they are 

 relatively triploid (e.g., in Crepis and Nicotiana rustica-paniculata) . 



