It is to be understood that not every individual represented by 

 C will be doubled even if A and B are precisely the same size. In 

 the first generation the average number of C individuals of maxi- 

 mum size will be 12 out of 100 under conditions stated. The 

 remaining 88 in 100 will all be larger than either Aor B and will 

 range in sizes between the size of parents and the maximum size 

 of offspring illustrated. As for the second and later generations 

 derived from C, since the A and B of subsequent fertilizations are 

 more closely related than were the original A and B, the percent- 

 age of maximum results runs up as high as 40 per centum, as I 

 found in my Petunia crossings. This will happen only in case of 

 close relationship between A and B. 



The third: 



Figure 3 illustrates the result when the ovule parent B is one- 

 half the size of the pollen-bearing parent A. In this case the 

 offspring C will be slightly larger than the pollen parent A, but 

 no great increase in size can be expected from this method. 



By observing the results obtained under the conditions repre- 

 sented by Figures i, 2, 3, one can determine exactly what to 

 expect out of material on hand, whatever the relative size of the 

 plants A and B may be. Size comparisons are made between parent 

 blooms when pollinating for the purpose of increasing the size 

 of blossoms; between fruits by increasing the size of fruit, and 

 between kernels by increasing the size of kernels. 



MODIFICATIONS OF COLOR 

 OR FORM OF FLOWERS 



The pollen-bearing parent is always the dominating factor in 

 changes of form or color. For instance, if one desires to increase 

 and accentuate incipient ruffling or frilling which may occur in 

 the petals of a given plant, A and B parents already having some 

 marks of the nature desired should be chosen; but the pollen 

 should be taken from the individual which shows the desired 

 feature most strongly. 



