1914] 



GATES — SOME CENOTHBRAS FROM CHESHIRE AND LANCASHIRE 389 



both the broad or normal type (pi. 20 fig. 5) and the elliptica 

 variety (pi. 20 fig. 4). The latter had a number of flowers with 

 elliptical petals and it also had a different method of branching. 

 Plate 21 fig. 12 is representative of a uniform F4 culture of 49 

 plants of the variety elliptica. This photograph is taken on a 

 larger scale, and the nodding of the stem is merely due to wilting. 

 This differs from typica (pi. 20 fig. 5) constantly in having nar- 

 rower leaves and short branches, as well as in the occasional 

 elliptical flowers which appear to be largely under environ- 

 mental control. 



The variabiUty of this race is therefore as interesting as are 

 the features, such as the general bud and leaf characters, in 

 which it is constant. The fact should also be mentioned that a 

 Zofa-hke mutant, doubtless having 15 chromosomes, appeared in 

 the F4 generation, and also a mutant resembling CE. mut. albida. 



CE. RUBEINEEVOIDES 



This race resembles (E. mut. rubrinervis in many features, 

 and yet differs from it constantly throughout. I have pre- 

 viously referred to this Birkenhead race as No. 25 (Gates, '11, 

 p. 350) and studied the variation of the red stripes on the buds. 

 In all, 1968 plants of this race have been grown in the years 

 1909-1912, so that four generations of offspring from a single 

 individual have been cultivated. An illustration of that indi- 

 vidual has aheady been published (Gates, '12, pi. 3). One fam- 

 ily of offspring was grown in 1909, two in 1910, eight in 1911 and 

 nine in 1912. Usually the variabihty of famihes progressively 

 decreased, since each family was derived from the selfing of 

 one individual of the previous generation. The discussion of 

 the precise ancestry of this race is of course out of the question, 

 but its characters bear nearly though not quite the same rela- 

 tion to the (E. Larharckiana from this region that the Lamarck- 

 iana and rubrinervis of de Vries's cultures bear to each other. 



The 1909 family, or Fi, numbered 111 plants. Plate 21 fig. 8 

 shows one of these as a rosette. The leaves are narrower and 

 more pointed than in mut. rubrinervis, and nearly smooth. 

 About 20 of the plants in this culture omitted the rosette stage 

 altogether and shot up a stem directly from the seedhng stage 

 (pi. 20 fig. 7). A normal mature plant of this family is shown 



