246 Inheritance m the Qroundsel 



Nos. actually found, Nos. calculated, 

 distributed amongst distributed amongst 



Totals ... 1351 387 315 649 337 337 675 



The radiate groundsel of the Channel Islands. Mr Marquand, to 

 whom living specimens of erectus, radiatus were forwarded, with a 

 request that he would send back the corresponding Channel Islands 

 form, obliged me by forwarding in April, 1907, several living specimens 

 in fruit. These were, as he also recognised at that time, quite distinct 

 from the Glamorgan erectus, radiatus. The ligulate ray florets (see 

 PI. XVI, fig. 10 and text-figure 30, p. 274) are not only shorter and so 

 less conspicuous, but more distinctly 3-toothed at the apex. Moreover, 

 the plants were nearly covered with a somewhat coarse type of hair, and 

 hence distinguishable by that character alone from all the preceding 

 types, which are glabrous. The plants had obviously been obtained 

 from a sandy habitat and consequently it was anticipated that there 

 would be some difficulty in raising normal colonies in the heavy soil of 

 my garden. Eayps. 17 and 18. Cultures were started on the day of 

 the arrival of the plants, and at a later date as well, and the parent 

 plants preserved for comparison. The plants raised on the Lias clays 

 of Glamorgan differ in many particulars from the parent plants, but the 

 RR character remains unchanged and the hairiness becomes so pro- 

 nounced as to form a whitish felt sufficiently dense to make the plants 

 conspicuous at a distance and to prevent, near at hand, the easy recog- 

 nition of the stem-colour (PI. XV, fig. 2, PI. XVIII, fig. 24). Pedigree 

 cultures have been continued, not without difficulty, for four generations, 

 and the plants adhere strictly to one type. They are always relatively 

 unhealthy, late (re(}uiring about 110 days to mature seeds, \vhi\e praecox 

 requires only 72 days), and demand extra care in selfing. Unselfed 

 plants may produce 20°/^ of hybrids. This constant hairy, radiate type 

 has been named lanuginosus. 



It seemed desirable to test once more the inheritance of the RR 

 character (and also of course its relations to other characters) in the 

 hybrids between lanuginosus on the one hand and praecox, erectus and 

 multicaulis on the other. It was certainly advisable to produce, if 



