W. E. Agar 319 



number of factorial differences, and hence more refined methods for de- 

 tecting segregation are required. In such cases, the pure parental com- 

 binations can only be expected to appear very rarely among large numbers 

 of descendants of the hybrid, and segregation is most readily detected by 

 an increase in the range and amount of variation in F2 and subsequent 

 generations as compared with F,. If the parents are homozygous, any 

 diversity in F^ depends upon purely non-genetic factors acting on 

 genetically homogeneous material, while. in the later generations, though 

 the non-genetic causes of diversity remain the same, the material is now 

 geneticall}'- heterogeneous, and hence the total amount of diversity is 

 increased. If the number of factors is so large that all the possible com- 

 binations are only likely to appear when the total number of individuals 

 is very great, the range of variation in the population found by the 

 descendants of the hybrid will continue to increase until the extreme 

 combinations have made their appearance. Moreover, while any indi- 

 vidual or pair of individuals from the F^^ generation should give offspring 

 not differing, except within the limits of error of random sampling, from 

 those of any other ^1 individuals, the groups of offspring of different 

 individuals of the F^ generation will differ from each other. In general, 

 F2 parents will produce F^ groups into a mean near to that part of the 

 range of variation to which they themselves belong. In other words, 

 there should be a correlation between parent and offspring in all genera- 

 ^tions later than F^. 



Examples of this method of detecting segregation among the factors 

 co-operating to give the distinctive features to complex somatic characters 

 are now among the common places of genetics, and the same methods 

 can be applied to a hybrid reproducing itself parthenogenetically. In- 

 deed, segregation must be much more easily detected in such a case, 

 since the results of segregation cannot be neutralized by recombination 

 of the segregated factors in syngamy. 



The material yielded by the clone Xc has been tested in both these 

 ways. The test of variability, as measured by the standard deviation and 

 range of variation has been applied to the males, both because this is the 

 simplest test, and because their sterility excludes the direct application 

 of the correlation test. This test has however been applied to the females. 



The relevant data regarding the inheritance of the -^ ratio in the 

 males of the hybrid clone are summarised in Table VI, which also 



