ON SUCCESSIVE DUPLICATE MUTATIONS. 213 



black (B) : 22 white (B). F3 offspring were grown from each of 

 these 108 plants and the results showed their composition to 

 have been as follows: 36 BB:5oBb:22 bb. From this result 

 Nilsson-Ehle concludes that there was a preponderance of 

 "black " gametes over "white " ones. But a series of F3 families 

 in another cross gave the reverse condition, 26 BB:6oBb: 33 bb, 

 from which the conclusion is drawn that white gametes were 

 here more numerous than black ones. Even though these re- 

 sults offset each other yet they cannot be referred merely 

 to chance fluctuations in ratios. But no further explanation 

 of them was offered. It will be shown later that in my crosses 

 of CE. rubricalyx and CE. grandiflora these deviating ratios do not 

 offset each other but are all consistent with the hypothesis 

 that R gametes are being produced with greater frequency 

 than r gametes. 



Returning now to Table II. the first ratio (68:16) is not a very 

 bad fit for 3:1, although exceedingly close to 4:1. It might 

 easily pass for 3:1 without further comment were it not for the 

 fact that two other ratios in this table are in very close agree- 

 ment with 4:1 while they depart very widely from 3:1. The 

 sum of these two ratios (468:112) is very close to 4:1 while it is 

 highly improbable as a 3:1 ratio, the more so since the actual 

 3:1 ratios are almost in precise agreement with expectation. 



Among the three F2 families from CE. grandiflora X rubricalyx, 

 two show a 15 :i ratio and one a 4:1 or perhaps a 3 :i ratio. Hence 

 it might be supposed that the rubricalyx plant which was used 

 as pollen parent, already possessed duplicate factors for red. 

 But this was not the case. That plant was in fact heterozygous 

 for a single factor, since when crossed with grandiflora it gave 

 an Fi of 79R : yir, which is as near to equality as could be ex- 

 pected. The full history of the rubricalyx individuals used for 

 this and the reciprocal cross has been given in pedigree form in 

 another paper, 1 to which reference should be made. It may be 

 said that in both cases they are descended from the family which 

 contained 33R : iir. One member of this family waspolHnated 

 by nanella and produced a family of 42 plants. One of the latter 

 (No. IV., 2), which was a perfect rubricalyx in appearance but 



1 Gates, Zeilschr. f. Abst. ». Vererb., 11 : opp. p. 216 and on p. 217. 



