242 Gates. 



drops to 1.48 : i, and the latter crossed back in the same way (culture 67) 

 gives only 1.23 : i. This reduction in 7?'s is all the more striking because 

 Nos. IX. 4 and VI. i crossed together (culture 76) gives the highest 

 ratio of all, namely, 5.86 : i. It thus appears that pure grandiflora when 

 used to cross back with the hybrid, exerts an inhibiting effect as regards 

 the development of R, reducing the percentage of individuals in which 

 R appears. That grandiflora also exerts an inhibitory effect in reducing 

 the quantity of anthocyanin pigment produced in an individual, I shall 

 show later. 



Plant No. IV. 2 from culture — (ruhricalyx :■: grandiflora) occurs 



in three 1912 families. In culture -- it yields almost exactly a 3 : i ratio ; 

 in culture 75, when crossed as pollen parent with its reciprocal, it pro- 

 duces a ratio approximating 3:1, but somewhat lower, the seed-parent 

 (No. XII. i) being evidently homozygous for R. A sensible reduction in 

 the percentage of R's in the cross may be accounted for by the fact that 

 grandiflora (in a "diluted" form) has been introduced from both parents. 

 Thus far the results are all consistent with the interpretation given 

 above. Let us now compare the results of the double-reciprocal crosses, 

 sections V and VII of table III. In culture 72, when the same plant 

 (No. IV. 2) mentioned in the last paragraph, is crossed as seed-parent, 

 with its reciprocal (No. XVI. 2), which was presumably homozygous 

 for R{}), we get an unexpectedly great increase in the 7?'s, which appear 

 in the ratio 11. 4 : i. This result alone is not explicable on the principle 

 that "more grandiflora^^ in the parentage decreases the percentage of 

 i?'s. The explanation of the fact that (ruhricalyx x grandiflora ) x 

 (grandiflora x rubricalyx) gives a different ratio from (grandiflora x 

 ruhricalyx) x (ruhricalyx x grandiflora ) is to be found in another 

 principle. Of these double-reciprocal crosses, the former (Cult. 72), 

 according to de Vkies's results, should most resemble ruhricalyx in 

 its general characters, while the latter should be nearer grandiflora. The 

 discussion of the foliage and other characters is deferred to a future 

 section [vide infra, p. 253), but it may be pointed out here that if this 

 principle is true it will account for the fact that the ratio is not depressed 

 below the high value 11. 4 : i, for along with the grandiflora characters 

 goes a depression in the ratio. In the other double-reciprocal cross 

 (Cults. 75, 76, 77), in which the .effect of the grandiflora grand-parents 



(1) Owing the limits of space and other causes it is impossible to carry out all 

 the collateral experiments which are theoretically desirable. Thus the offspring of 

 plants No. XVI. 2 and XII. i from self-polhnation have not been grown, but it is 

 quite safe to assume from the olher evidence that both were homozygous for R. 



