THE POLYPLOID SERIES IN OSMUNDA 



if they occur in suitable positions for chiasmata to form in relation to each of the 

 homologues, will result in the estabUshment of a union between all three which is 

 retained until metaphase with the production of the characteristic shapes associated 

 with trivalents (Fig. 22c). If the chiasmata are unsuitably placed the group will fall 

 apart. 



The analysis of the triploid nucleus of Fig. 22 c is given diagrammatically in Fig. 23, 

 in which trivalents are shown in black and pairs and univalents in outline. Similar 

 analyses for 109 cells are summarized in Table 2, and it is important to notice that 

 although every chromosome is known to be present in triplicate, a circumstance which 

 might lead one to expect that 22 trivalents would always be formed, the actual numbers 



1^ 



«• 



ifi 





^ 



Q 





^ 



Triploid Osmunda 3n - 66 



Fig. 23. Explanatory diagram to Fig. 22c 

 with trivalents in black but pairs and 

 univalents in outline. 





^@ 



ea 





4^ ^^ -to 



85 



Tetraploid Osmunda -^/7 = 88 



Fig. 24. Explanatory diagram to Fig. 22 rf 

 showing quadrivalents in black and pairs in 

 outline. 



found show a random distribution round an ill-defined mean which is considerably 

 less than this (i.e. between 14 and 19). The observations recorded in the table are from 

 several plants in different years, a fact to which the absence of a well-defined peak in 

 the distribution is probably due, since the average almost certainly varies slightly 

 with external conditions from year to year. Had a larger sample been analysed, 

 the range of actual numbers would probably have been slightly wider, and it is 

 possible that in extremely rare cases the maximum of 22 trivalents are actually formed. 

 Such further possibilities are, however, of no importance from the present standpoint, 

 for the main purpose of Table 2 is to indicate that, had the origin and nature of the 

 plants been unknown, it would have been safe to infer autopolyploidy if the average 

 number of trivalents found could be shown to involve more than half the chromosomes 

 present. This fact has, of course, already been used for the analysis of the Biscutella 

 series noted in Chapter i . 



37 



