The Experimental Polyploids 281 



ities of classification become very real. These are problems requiring 

 iurther study which cannot be resolved entirely in this review. There 

 are other cases. In fact, the gioup between the autoploid and amphi- 

 ploid provides the most interest and perhaps the greatest opportunity 

 lor practical and theoretical work in polyploidy. E\en though one 

 cannot decide definitely on the classification, there is no need for 

 concern, for he may utilize the opportunities presented by these 

 intergrading polyploids without classifying them. 



One way to explore this group has been oj^cned by an inquiry into 

 the special kind of polyploid called the "segmental allopolyploid."*''^ 

 Good reasons were given to justify the establishment of this special 

 group. Some types of polyploids have segments of chromosomes so 

 closely associated that pairing is between the two parental genomes, 

 and therefore they cannot be considered as strictly amphiploid; but 

 in other segments, there is enough differentiation to prevent pairing 

 of the chromosomes that originate from the different parents. View- 

 ing the chromosomes segment by segment, instead of as whole chromo- 

 somes or even whole genomes, gives one a more critical picture of the 

 basis for borderline types between the autoploid and the amj^hijiloid. 

 Theoretical and practical aspects are greatest among the ])olyploids 

 that fall between the unquestionable autoploid and amphiploid. 



Pairing of chromosomes is of limited value in classifying the 

 polyploids e\'en though this cytological method is one way to point 

 out the difference between the autoploid and the amphiploid. Some 

 diploid species hybrids may show pairing at the diploid level, but 

 this does not necessarilv happen. On the other hand, complete lack 

 of pairing at the diploid level does not insure total bivalents at the 

 polyploid stage.^- Less and less reliability is being placed on pairing 

 of chromosomes as a measure of homology and a means of distinguish- 

 ing the autoploid from the amphiploid. As more examples come into 

 view, the case for pairing is increasingly complicated. Other factors 

 must be considered. 



Sterility and fertility characteristics may separate the amphiploid 

 from the autoploid. The latter is invariably less fertile than the 

 diploid, and the amphiploid changes from a sterile condition to a 

 fertile one upon doubling of the chromosomes. In reviewing many 

 cases, one can find wide variation in degree of sterility among the 

 autoploid and the amjjhiploid cases. Actually, the causes of sterility 

 are so complex that this relationship is of little help in trying to 

 classify the two types. Yet basically, sterility may be closely related 

 to some basic cytogenetic mechanism. 



The best solution to the classification problem appears to he the 

 chart developed bv Cilauscn and his colleagues^'' on which they place 

 the amphiploids in a relative position depending upon a series of 



