E. R Saunders 349 



were actually observed. The agreement between the observed and 

 calculated results is so close that we may conclude that the same 

 gametic series is formed by the reciprocal cross-breds {xyw % x XYW ^) 

 of the present mating and {XYW % x xyw </) of mating 3. If n has 

 the same value in each of the F^ plants, and if in the present mating 

 the non-double-throwing parent is homozygous in XYW, we must 

 suppose that the absence of the two rarer forms in the two families 

 mentioned above is a chance variation. 



Mating 4. c?-non-cream ? x no-d-cxeaxo. ^, Not yet carried out. 



Mating lO. Reciprocal cross, no-d-cream % x d-non-cream ^. 



Summary of results. For details see Table IV. 



Five kinds of matings of this type were carried out, viz., 



7io-(i-glabrous cream % x d-hoary white (intermediate) (/". 



„ „ ? X c?-glabrous white ^. 



„ „ $ X d-glabrous red ^. 



„ „ $ X c?-glabrous azure ^f. 



„ „ $ X c?-glabrous light purple f^. 



72 .F, families were raised, all of which included some doubles, the 

 proportion varying from 3 s. : 1 d. to a very much higher proportion of 

 singles. A point of special interest in this group of matings is that no 

 doubles with cream plastids were recorded in an F^ generation numbering 

 more than 3000. That is to say, in no case in which a non-double- 

 throwing cream has been employed in a mating with an eversporting 

 non-cream form has it yet been found possible to obtain the com- 

 bination of creamness with doubleness in F^, though a considerable 

 number of the single F^ plants have cream plastids. The total numbers 

 obtained were : 



1666 singles with white plastids 



773 doubles „ „ „ 



790 singles „ cream „ 



or about twice as many of the form with both dominant characters as 

 of either of those exhibiting one dominant and one recessive character. 

 In matings of this type F^ is presumably derived from the union 

 o{ XYW ovules with xyW pollen. Now if the two kinds of germ cells 

 which united to produce F^ were formed again by F^ without any 

 redistribution of the factors for plastid colour and for singleness and 



^ See note, p. 326. 



