Breeding experiments which show that hybridisation and mutation etc. 2^0 



(from plants in bloom) was 34 : 24, while in the reciprocal cross this 

 ratio was 30 : 28. The latter ratio is practically identical with, and 

 the former sufficiently close to, the anticipated 1:1. A fact equally 

 important is the exact 3 : i ratio (33 : 11) into which the family (see 

 pedigree i, p. 216), from which the ruhricalyx parent of both reciprocal 

 crosses was derived, was found to split. Taken together, these three 

 ratios show clearly according to Mendelian interpretation, that a 

 single unit-character difference is present in this ruhricalyx family, 

 and that the ruhricalyx parent of both reciprocal crosses was hetero- 

 zygous, producing R and r germ cells in equal numbers. This con- 

 clusion is also in harmony with the origin of ruhricalyx by a single 

 germinal change, and we must therefore accept this Mendelian evi- 

 dence as the basis of our interpretation. We should therefore anti- 

 cipate that the 3 : i ratio would continue to recur in the F» generation, 

 but this is far from being the case. 



Unfortunately, owing to the fact that a number of the cultures 

 in 1 9 10 and 1911 remained rosettes, the collateral families given in 

 pedigree i offer little evidence regarding the ratio between R and r. 

 But in three 1911 families (from Nos. IV. 3, IV. 9 and III. 10), the 

 ratios (36 14, 22 : 23. 30 : 19) as determined from the rosettes only, 

 appear to depart widely from the expected 3:1. 



An examination of table III reveals at once, in the first three F2 

 cultures from grandiflora ■ ruhricalyx, a remarkable range in the 

 R : r ratio. Culture 48 gives about 4:1, culture 49 nearly 10 : i, and 

 culture 50 only 4 r to 133 R. I am inclined to attribute these 4 /s 

 in the last family to reversions, in which the dominant character sud- 

 denly acquired has been equally suddenly lost again. In pure (homo- 

 zygous) ruhricalyx, of which I grew a culture of 200 plants last year, 

 not a single one reverted to ruhrincrvis, nor could I discover any 

 marked variability in the amount of pigmentation. The 4 plants 

 above-mentioned I believe to be reversions in a race which was homo- 

 zygous for R, caused by the disturbance resulting from crossing. 



If we now examine the seven F2 families from ruhricalyx x grandi- 

 flora (II A. table III) we find that four of them (marked a) yielded 

 either exactly of approximately a 5 : i ratio; two of them (marked h) 

 yielded almost a perfect 3 : i ratio; while in another (marked c) the 

 ratio was somewhere between 19 : i and 4 : i (see footnote, table III, 

 p. 236). It is obvious that the ratios in the two groups of families 

 a and b are significantly different from each other. Another fact of 

 much importance was not discovered until the pedigree numbers of 



