132 



University of California Publications in Agricultural Sciences [Vol. 4 



samples of seed were sorted, and the smaller and larger seeds planted 

 separately. 



Table 31 gives the data from this test. Here is practically con- 

 clusive evidence (see tables 39 and 40) that the smaller seeds much 

 more often contain embryos of the crenate type.^* Since the embryo 

 of a Matthiola seed occupies practically all the space within the seed 

 coats, it is evident that even as embryos Snovvflake plants exceed 



Table 31 



Cultures of 1913. Crenate-leaved type: proportions from smaller and larger 



seeds of crenate parents. 



Respectively 64.6 and 81.8 per cent of the numbers of seeds planted. 



crenate plants in size. This fact, obviously, is further evidence in 

 favor of the hypothesis of partial selective elimination of crenate 

 heterozygotes during embryonic development. 



It may be worth noting that the 73 plants from the smaller seeds 

 include 6 (8) apparent mutants of other types (mutation coefficient 

 11.0 per cent), while the 193 plants from the larger seeds include 

 only 3 apparent mutants (1.6 per cent). 



Before we can profitably discuss these data further, we must con- 

 sider the results from cross pollination (tables 32 and 33). The 

 numbers, though small, make it very probable that both eggs and sperms 

 carry the crenate factor. Further, it appears from series 20 that only 

 a small portion of the sperms carry this factor, as we should expect 

 from its apparent linkage with singleness. If homozygotes are non- 

 viable, the combined crenate percentages of reciprocal crosses should 



'^ The poorer germination of the smaller seeds suggests that the disparity 

 between the two lots of seeds in the proportion of crenate embryos was even 

 greater than the cultures indicate. 



