378 MEMOIRS OF THE NEW YORK BOTANICAL GARDEN 



Here as in the F2 the inheritance of flower color is evidence that 

 seed-setting is the result of self-fertility and is not due to partheno- 

 genesis or to stray pollinations resulting from experimental error. 

 Six of the series were from Fo white-flowered plants and all of these 

 were composed of white-flowered plants, although it should be 

 stated that there was considerable variation in respect to the 

 development of blue stripes on the anthers. The fact that F2 

 white-flowered plants gave only white-flowered progeny is evidence 

 that the seed from which they grew was not the result of acci- 

 dental cross-pollination. The five other series of the F3 were 

 from F2 blue-flowered plants, and, with one exception (which is 

 series {E22 X A)-io-6- of only two plants), the progeny split 

 up, giving a total of 33 blue-flowered and 8 white-flowered plants. 

 This we may also consider as evidence of self-fertilization. 



Summary of the Fi, F2, and Fz generations. — For purposes of 

 summary and comparison the data for the three generations and 

 for certain families are presented in table 6. Here the frequency 

 distribution is given with the plants grouped in classes on the 

 basis of their self-fertility; the average fertility for the self- 

 fertile plants and for the entire generation or series is computed 

 from the actual percentages already given in tables 3, 4, and 5. 

 One of the F2, and one of the F3 self-fertile plants are not included 

 in this distribution because data on the seed number are not 

 complete due to the gathering of seed by birds. It is fully recog- 

 nized that a larger number of pollinations might change the 

 results in numerous cases, that plants which appear self-sterile 

 may in reality be but very feebly fertile, and that some of the 

 seed-setting in cases of very low fertility may be due to stray 

 pollinations. The number of plants in the F2 and the F3 genera- 

 tion is nearly the same, and the methods of manipulation used 

 were quite identical, so it would appear that such a comparison 

 can quite properly be made. 



In the first generation, 15 plants out of 172 were self-fertile in 

 some degree; the average percentage for the 14 plants upon which 

 flower data are available was 11. 6, and the percentage for the 

 entire generation was 0.94. The F2 whicli w^as grown from self- 

 fertile parents and in which selection for self-fertility first appeared 

 gave 39 self- fertile plants and 41 self-sterile jilants; the average 

 fertility (of 38 plants as figured) is 21.9 \)ev cent and that of the 



