A. H. Trow 241 



certainly flourishes in localities where S. squalidus does not occur, and 

 as we shall see, these two species appear to have no necessary association 

 with each other. 



A capitulum of this radiate groundsel, from one of 49 plants 

 of the pure strain cultivated in 1912 (7th Generation) which I have 

 named erectus, radiatus, is represented in PI. XVI, fig. 8. This strain 

 occurred originally as a weed in my own garden along with a non-radiate 

 form, indistinguishable from it until the opening of the flowers, when 

 of course the difference is very obvious. The non -radiate type may 

 be styled erectus. Its capitulum is shewn in PI. XVII, fig. 14. Either 

 form may thus be regarded as a true variety of the other, using the 

 term variety in the sense suggested by De Vries. 



In the year 1904, I noticed a single plant whose capitula differed 

 from those of the two preceding. Each capitulum had eight to 13 

 ligulate ray florets, but the ligules were only one-half the usual length 

 and were conspicuously three-toothed at the apex (PI. XVI, fig. 5 and 

 text-figure 29, p. 274). Perhaps influenced by the prevalent view that 

 groundsels are autogamous plants (self-pollinated), this was regarded 

 as a mutation, and the supposed mutant was allowed to seed freely, 

 but no seeds were collected. In 1905, as a result of the natural sowing, 

 a small crop of groundsels appeared in which the three types were 

 present, and I resolved to start a set of pedigree cultures to determine 

 their relationships to each other. 



It will be convenient to denote the three types as follows : — (1) the 

 radiate plants with the ray character fully developed by RR (PI. XVI, 

 fig. 4), (2) the non-radiate plants by NN (PI. XVI, fig. 6), and (3) the 

 intermediate plants with the ray character partly developed by NR 

 (PI. XVI, fig. 5). The figures apply equally well to the erectus and 

 multicaulis types. 



In all the experiments methods were employed to secure the absence 

 of all foreign seed from the cultures. The soil and seed-pans were 

 sterilized by submission to one hour's steaming. So effective were these 

 precautions that from 1905 to 1911 I have only found two invading 

 seedlings in the seed-pans ; one of a THfolium, the other of Poa annua, 

 while the controls were invariably sterile. 



Preliminary experiments, Nos. 1, 2, 3, 4 and 5. The first experi- 

 ments were more or less tentative and were made to test the inheritance 

 of the full ray-character in RR plants. Exp. 1. A single head of seeds 

 (really fruits) was collected from a plant growing as a weed in the 

 garden in September, 1905, and the seeds were sown at once. Thirty- 



