270 Inlierltanee of Gltitne Length and of Colour la Oats 



eminently unsound. Additional data are needed. Further breeding 

 on of all the colour-yielding forni.s in the cross furnishes us with them; 

 but, as will be seen from Table IV, into which they have been gathered, 

 although of precise and distinctive kinds, by their very multiplicity 

 they tend rather to complicate than to clarify the problem. 



For the sake of convenience letters have been used to denote the 

 various glume and colour types in the table. Whenever the parental 

 colour is enclosed in brackets, the identity of the F-^ segregation with 

 others of definitely known parentage has been the means of its deter- 

 mination. 



In the first part of Table IV are the flushed forms which bred true. 

 They are followed by one which gave rise to 18 flusheds and 6 streakeds. 

 The possibility that this may really belong to the next group, which 

 posse.sses non-coloured elements besides and conforms to the ratio 

 VIF : 38 : IN, is admitted ; but other e.\amples of the same sort of 

 .segregation (picked out of F^ results and therefore not included here) 

 show that in any case a type throwing ^F : IS does exist. Some 

 members of the 12 : 3 : 1 section, perhaps, by reason of their incom- 

 pleteness will call for a few words of explanation. No. 4 has 17 flusheds 

 and 4 streakeds but no non-coloured. Although it is quite conceivable 

 that it may belong to the 3F : 1»S' class, the low proportion of streakeds 

 gives cause for hesitation in so placing it. Unfortunately only one of 

 the 17 flusheds was carried on for another generation; but as it yielded 

 15 flusheds and 3 streakeds — again a low proportion of the latter — 

 with one of the streakeds very faint indeed and very likely hetero- 

 zygous, it was decided to retain No. 4 where originally placed. In 

 No. 3 there were no streakeds. Out of four of the flusheds which were 

 grown on for another year, however, one gave rise to 14 flusheds and 

 6 streakeds. The other three bred true. If it were a case of two 

 cumulative factors making for the 'flush' character, and no 'streak' 

 factor in the zj-gote at all, this type of segregation could not have been 

 produced. A\o flushed : 1 non-coloured ratio must therefore be rejected. 

 Moreover, as the single non-coloured forms less than one quarter of the 

 total. No. 3 goes to the 12F : 3S : IN group instead of to the following 

 set, wherein the non-coloureds are produced in considerably greater 

 relative numbers and the streakeds constitute the smallest term in the 

 ratio 27 : 9 : 28. By analogy No. 7 does likewise. 



Two more groups bred from flushed parents remain. In neither do 

 any streakeds appear. The first has flusheds to non-coloureds in the 

 ratio 3:1, the second !> : 7. 



