220 Gates. 



tinued yielded an exact 3 : i ratio (33 : 11) of the dominant and recessive 

 types. Another self-polhnated flower from the same plant gave in 

 1910, 70 rosettes. Fifteen of these came into bloom, of which 10 were 

 rubricalyx and 5 rubrinervis. A sister plant (070.32) of the same family 

 yielded 13 rubricalyx and 7 rubrinervis, nearly all of which came into 

 flower. 



Of the 33 plants in the culture above-mentioned, the five which 

 were tested all proved to be heterozygous (see pedigree i), but since 

 the numbers were small and the plants all remained rosettes, in which 

 there is some uncertainty of identification, the ratios obtained cannot 

 be regarded as significant. One of the heterozygous plants (No. IV. 5) 

 was crossed with 0. Lamarckiana, and another (No. IV. 6) — which is in 

 our main pedigree of descent — with the dwarf nanella. The latter cross 

 should be compared with a similar one (0. rubrinervis x nanella) of 

 DE Vries (1903, II, p. 451) with which it agrees. The Fj contained 

 42 plants, of which 7 which did not bloom were Lamarckiana, and the 

 remaining 35 were rubricalyx or rubrinervis. Twelve of these, which 

 bloomed, were all rubricalyx. Two of the latter when tested proved to 

 be heterozygous; a third (No. III. 10) when tested yielded 49 rosettes of 

 rubricalyx or rubrinervis, 3 dwarf rosettes, i aberrant mutant, and 

 6 Lamarckiana rosettes, some of which were, however, more or less 

 intermediate betwen Lamarckiana and rubrinervis, and four of which 

 had in addition the red pigmentation of rubricalyx. A fourth plant 

 (No. II. 7) in this family was used to pollinate Lamarckiana; and a 

 fifth (No. IV. 2) was the male parent of the cross grandiflora x 

 rubricaly X . 



Another collateral pedigree shown in pedigree i is derived from 

 the II rubrinervis (rr) plants segregated in the second generation 

 from rubricalyx. One plant (No. II. 3) selfed gave 39 rubrinervis. Two 

 others (No. I. 2 and III. 5) were used to pollinate Lamarckiana, yielding 

 in both cases, as expected, an Fj containing Lamarckiana and rubrinervis, 

 the totals being 29 Lamarckiana to 67 rubrinervis. No. III. 5 was 

 also pollinated from Lamarckiana and yielded in F^ 12 Lamarckiana 

 and 54 rubrinervis. Thus in these reciprocal crosses the frequency 

 of rubrinervis always shows a certain (variable) preponderance, as 

 DE Vries has also observed. 



0. nanella was also pollinated with rubrinervis No. III. 5 and 

 in Fl gave 25 Lamarckiana and 52 rubrinervis. One of the latter when 

 tested in 191 2 gave 10 nanella and 32 rubrinervis, in agreement with 

 previous results of de Vries and of myself. (See pedigree i.) 



