INFERENCE OF POLYPLOIDY 197 



(h) The chromosomes of the haploid set must be numerous. 



Crepis iectorum with 4 chromosomes gives a relatively fertile 

 tetraploid because the chance of unbalance is reduced. 



(c) There is a correlation between the genetic differentiation of 

 the chromosomes of species and their structural differentiation. 

 In other words, structural changes have always occurred along with 

 other genetic changes and prevent the pairing of the chromosomes 

 in hybridity. This follows from our earlier conclusion that all inter- 

 specific hybrids are also structural hybrids. 



{d) The chromosomes of each set must be qualitatively 

 differentiated in a fairly high degree. This third condition is not 

 applicable to high polyploids (such as Rumex Hydrolapathum, 

 2n = 200, Kihara and Ono, 1926 ; and Prunus Laurocerasus , 

 2n = 180, Meurman, 1929 a), nor apparently to Hyacinthus 

 orientalis, for this species combines high fertility of the triploid with 

 viability of unbalanced forms and lack of differentiation amongst 

 them (D., 1929 h). Its chromosomes are therefore not qualitatively 

 differentiated within the set (Ch. VIII). 



3. POLYPLOID SPECIES 



In about half the species of the angiosperms the gametic 

 chromosome number is a multiple of that found in some related 

 species, the chromosomes being themselves comparable in the two 

 forms. From this alone it is clear that they owe their origin to 

 polyploidy, tetraploids being derived from crosses between diploids 

 and diploids, hexaploids from crosses between tetraploids and 

 diploids, and so on. They may therefore be expected to resemble 

 the polyploids produced in experiment. 



Winge (1917), in a remarkable survey of these conditions, first 

 suggested that they were likely to be of the type derived from 

 hybridisation (allopolyploids). He considered that the need for a 

 partner with which a chromosome could pair would stimulate 

 doubling in hybrids. This teleological argument led to a sound 

 general conclusion, although the argument itself is admissible only 

 as a figure of speech. Somatic doubling probably occurs no more 

 freely in haploids (in which the chromosomes have no regular 

 partner) and in hybrids (in which they have no effective partners) 



