242 Inheritance in the Groundsel 



six seedlings were raised, eight of these ultimately flowered, and all 

 proved to be RR. Exp. 2. One of these eight plants was isolated in 

 the greenhouse, and from its seeds one hundred and nineteen seedlings 

 were raised in the following year. Of these one hundred and sixteen 

 flowered, and all were RR. Exp. 3. Another experiment carried out 

 under somewhat diflerent conditions gave the following result. Five 

 plants flowered in 1905, all RR. Exps. 4 and 5. From two of these 

 five plants seeds were obtained (a) after isolating in the greenhouse, 

 (6) after selfing in the open air. These seeds produced respectively 74 

 and 13 plants, all RR. Altogether 216 plants were raised in two 

 generations, and all were RR, breeding perfectly true to the parent 

 type. 



Exps. 6, 7 and 8. Further experiments were set up to test the 

 inheritance in all three types under natural conditions — cross polli- 

 nation by natural agents being possible. Three plants growing close 

 together, but not in contact, in the open garden, of types NN, RR and 

 NR were marked down for seed-saving on September 16th, 1905. 



Seed was collected as follows : — 



From NN. From Oct. 16th to Nov. 17th. 33 Heads on 17 separate days. 

 From RR. From Sept. 16th to Nov. 17th. 43 Heads on 20 separate days. 

 From NR. From Sept. 20th to Nov. 17th. 30 Heads on 14 separate days. 



The results are summarised in 



The interpretation of these three experiments is clear: NN and 

 RR are homozygotic forms so far as concerns the ray character, and NR 

 is the corresponding heterozygotic form. In addition to this main con- 

 clusion it may be noted that cross-fertilization, with the production of 

 natural hybrids, takes place in both ways, N x R and R x N, and may 

 produce an error of about 1 °/^ unless measures are taken to prevent it. 

 This error is so small that it might be neglected, if constant. Sub- 

 sequent experiments, however, shew that the error due to the absence 

 of selfing may reach 10 °/^ or more. Such free natural crossing as is 

 represented by a 10 "/^ error is due to (1) vicinism in its various forms, 

 (2) visits hy honey-bees — in a bad honey season, the groundsels, radiate 

 and non-radiate, are freely worked by these insects, (3) Aphis, the 



