82 THE PLANT WORLD 



purple to red, from red to white, and then to yellow. A 

 pure yellow, the original color of the plant, never produces 

 a color variation. To explain this, we have only to imagine 

 the germ cells of the purple chrysanthemum to confain the 

 factors for determining blue, red, white and yellow colors, 

 and as one or more of these colors are dropped, a different 

 color appears. It is probable that the yellow color is a 

 plastid color, (or at least belongs to a different pair) and 

 hence is recessiv^e to white ; while the other colors are sap 

 colors, and are dominant to white, although I hav^e no 

 evidence on this point. 



It is safe to say that in horticultural plants whose 

 flowers tend to vary in color, such as carnations, azaleas, 

 geraniums, roses, etc., the tendency is toward the loss of a 

 color factor. There are many cases in which a color is 

 deepened in a variety which already possessed the ability to 

 form the dye, but this is probably a different thing from 

 definite bud variation. 



Many instances illustrating the same same feature might 

 be given in the fruits : purple plums produce yellow plums, 

 blackberries produce white variations, red currants give white 

 varieties, and dark grapes throw white sports. In nearly all 

 cases there is a definite loss of a color character. Sometimes 

 this loss is permanent; while in other cases the color is not 

 able to develop, but is present in latent state, and may after 

 a longer or shorter lapse of time, again appear. But in 

 these cases the change is not a change in any direction, but 

 a loss or gain of a particularly loosely connected unit 

 character. 



A second common type of bud-variation, is the pro- 

 duction of cut leafed varieties in different species. In these 

 productions, the character which produces the development 

 of entire leaves is often retained in a latent state, and there 

 is frequently a jump back to the original form. The changes 

 taking place certainly affect unit characters, but there are few 

 data concerning their behavior on crossing. De Vries, how- 



