90 



THE AMERICAN NATURALIST [Vol. XLIV 



nearly always made between sibs in wholly white-flow- 

 ered families. Such crosses are necessarily homozygous 

 with respect to the absence of one (or both) of the two 

 genes whose joint action is necessary to the production 

 of color. Crossing of white-flowered individuals of dif- 

 ferent parentage or of white sibs in hybrid families will 

 doubtless quickly demonstrate the production of purple- 

 flowered offspring by white-flowered parents, in the ratios 

 required by theory. 



Summary 



The purple color in Lychnis dioica L. is a compound 

 character, produced by the interaction of three distinct 

 and independent genes in a manner exactly analogous 

 to the similar colors in Lathy rus, Matthiola, etc. 



The two types of purple color present in different in- 

 dividuals are a reddish and a more bluish-purple, the 

 former being changed to blue by treatment with alkalies, 

 and the latter changed to red by the addition of weak 

 acids. 



The bluish or alkaline color is hypostatic to the reddish 

 or acid color, this being the reverse of the condition found 

 in all other plants containing similar series of colors 

 which have thus far been reported, unless possibly an 

 isolated statement should prove correct that in Primula 

 Sinensis 1 1 blue is hypostatic to all the red shades." 



It is impossible to determine at present whether this 

 reversal of the relation between bluish and reddish 

 anthocyan results from the occurrence of positive char- 

 acters for both alkalinity and acidity, or whether only 

 one of these exists as a positive character and the alter- 

 native color is produced when this positive color-modifier 

 is in the heterozygous state, the latter situation involv- 

 ing the dominance of absence over presence. 



The rather wide fluctuation in the percentage of purple- 

 flowered families resulting from the cross of heterozy- 

 gous purple with white (i. e., supposedly DR X R), re- 

 ported in a former paper, may have been due in part to 



