216 HEREDITARY TRANSMISSION [CH. 



as in the primary hybrid, if the chromosomes bear the 

 " factors " for Mendelian characters, it is to be expected that 

 no segregation will occur. But FEDERLEY observed that 

 in the spermatocytes of the hybrids, at least in some cases, 

 one or two chromosomes paired and separated, and he 

 suggests that the segregation of the larval character is due 

 to its being borne by one of these chromosomes. 



The second example of direct evidence in favour of the 

 hypothesis is that of the late R. P. GREGORY on a tetraploid 

 race of Primula sinensis. In this race there is twice the 

 normal number of chromosomes; that is to say, there are 

 four instead of two of each kind. The race is fertile when 

 two tetraploid plants are crossed together; each germ-cell 

 contains the diploid number of chromosomes, and the zygote 

 again is tetraploid. If, now, the chromosomes determine 

 the segregation of Mendelian characters, such a tetraploid 

 race should show a new type of segregation, for in the zygote 

 there would be four chromosomes of each kind, and in the 

 germ-cell two. Suppose a plant to receive the character A 

 from one parent, and its allelomorph a from the other, then, 

 since each chromosome is duplicated, the heterozygote 

 would be AAaa, and instead of producing two kinds of germ- 

 cells, A- and ^-bearing, it would produce three, AA^ Aa, and 

 aa, in the ratio of I : 2 : I . Such a plant self-fertilised would 

 give fifteen offspring showing the dominant A character to 

 one showing the recessive a. This is exactly what GREGORY 

 found; instead of getting a 3 : I ratio when the heterozygote 

 is self-fertilised, he got ratios approaching 15 : I, and simi- 

 larly, when the heterozygote was crossed back with the 

 pure recessive, instead of a I : I ratio he got 3:1. At first 

 sight this appears a convincing demonstration that the 

 Mendelian characters of Primula sinensis are borne by the 

 chromosomes, or at least that their segregation is determined 



