R. RuGGLBs Gates 359 



condition, without reversion to either the ruhrinervis or the rubricalyx 

 type of pigmentation. Moreover, back-crosses of the original F^ hybrids 

 with rubricalyx intensify the pigmentation, while back-crossing with 

 [irandiflora dilutes and modifies it, producing a spotted condition of 

 the sepals in certain cases. I may perhaps be pardoned for main- 

 taining that all these facts confirm my view that, (1) the rubricalyx 

 pigmentation originated through a unit-change, as demonstrated by 

 the ratios before crossing and in the F^ hybrids; (2) this sharj) 

 unit-character may be modified and broken up by crossing with a 

 physiologicall\- diverse species. An adequate e.xplanation has not yet 

 been found for the F., ratios, such as .5:1 and 10:1, found in these 

 hybrids, but a little thought will show that any argument is fallacious 

 which attempts to argue backwards, as Shull has done, fi-om these ratios 

 to the original condition of the unit-character R before it was crossed. . 



In Shull's paper he devotes considerable attention to the red 

 spotting of the rosette leaves in his hybrids. This is a conspicuous 

 feature of Oe. grandifiora, and I have devoted much time to the study 

 of its inheritance in hybrids with rubricalyx. There is an evanescent 

 stage in the development of the rosettes of these F^. hybrids, when on 

 superficial examination it seems easy to classify them into red-spotted 

 and non-red-spotted. But more intensive study shows that every 

 degree of spotting exists, from rosettes in which all the leaves are 

 well spotted to those in which only one or two tiny sjoots appear on 

 the whole rosette. In other words, there is a continuous series from 

 rosettes with a large amount of spotting to those with none at all. 

 This feature was not found to be correlated with any well-marked 

 condition of the adult plants (as is the case with the ventral red of 

 rubricalyx rosettes), and as the classification of the rosettes is neces- 

 sarily more or less arbitrary and hence not dependable, it was not 

 considered worth while publishing ratios in which there was such a 

 large margin of eiTor. 



When Shull speaks of " negative correlation " between the pig- 

 mentation of the buds and of the stems and rosettes in his triple 

 hybrids, he is observing complex phenomena which have not been 

 analyzed. I have observed a large number of similar phenomena in 

 F.2, and F^ hybrids of grandiflora and rubricalyx, but have awaited 

 further data on the subject before publishing. It means, of course, 

 that red pigmentation in different parts of the same plant may be 

 separately inherited, as is known to be the case in various other plants. 

 Shull's " negative correlations " appear to depend, in part at least, upon 



