Breeding experiments which show that hybridisation and mutation etc. 243 



should be more conspicuous, we have, as aheady pointed out, a de- 

 pression of the R : r ratio to 3.24 : i, which is not significantly different 

 from 3:1. 



The first suggestion of Mendelians who examine these records will 

 be that R is after all not a single unit-character, but that two or more 

 independent "factors" for the same colour pattern are being independently 

 distributed in the germ cells. This type of hypothesis, based on no 

 evidence except the ratio obtained, has been adopted by a number 

 of Mendelian workers, and it was indeed suggested to me to explain 

 the variability in depth of colour, by one who was viewing some of 

 my experiments. 



Nilsson-Ehle (1909) formulated this hypothesis to explain the in- 

 heritance of red and white glumes in wheat, and I naturally attempted 

 to apph^ it to my results. The hj^pothesis has been widely adopted by 

 Mendelians, and has culminated in such absurdities as the suggestion 

 of Tammes (1911) that two or three "genes" (why not genii?) are requi- 

 red to open fully a capsule of flax ! 



The application of such an hypothesis to the inheritance of R in 

 rubricalyx crosses is, apart from the inherent improbability that orga- 

 nisms are constituted as such an hypothesis assumes, rendered im- 

 possible in this case by each of three lines of evidence: (i) The origin 

 of the character by a single sudden germinal change (vide supra, p. 217). 

 (2) The 3 : i ratio obtained in the Fo offspring of the mutant and the 

 approximately i : i ratio obtained in both cases in crossing reciprocally 

 with grandiflora, show that germ cells of the parent type and the new type 

 were being produced in equal numbers. (3) The ratios obtained in the 

 F2 of the crosses with grandiflora are impossible on the Mendelian as- 

 sumption of two or any other number of independent "factors" for red. 



If two independent factors for R are at work, the dihybrid ratio 

 9:3:3:1 becomes 15 : i, if three factors, the ratio of R to absence 

 of R becomes 63 : i. But the ratios in table III (p. 235) are so remote 

 from 15 : I, the nearest ratio, that any such theory falls totally to the 

 ground. The simple hypothesis of degrees of prepotency in the germ 

 cells of different individuals, while it may not have the appearance 

 of making the whole phenomenon of the inheritance of R plain and 

 easy, yet is at least in accord with all the actually observed facts. It 

 moreover assumes that the difference between R and r is fundamen- 

 tally quantitative and not qualitative. This is indeed, so far as I can 

 judge, its chief advantage over the idea of fixed units, for the term prepo- 

 tency in itself is of course not an explanation. 



i6* 



