No. 518] COLOR INHERITANCE 89 



percentage of purple-flowered individuals in families re- 

 sulting from crosses of purple with white was shown to 

 form a normal probable error curve, thus conforming 

 very well with the Mendelian hypothesis that gametes 

 of different alternative composition unite according to 

 the laws of chance. While this conclusion is not in any 

 way opposed by my later studies, it is now known that 

 a portion of that apparently fluctuating variation may 

 have been due to the occurrence of a mixture of sev- 

 eral different ratios. In accordance with the present 

 demonstration that the purple color is due to three fac- 

 tors, the combined action of two of which are necessary 

 to the production of any color and the addition of a third 

 for the modification of this color, a cross between white 

 and purple must give either all purple, or purple and 

 white in any of the following ratios : 3 : 1, 1 : 1, 3 : 5, or 1 : 3, 

 though the ratio 1 : 1 occurs much more frequently than 

 any of the other possibilities. In other words, without 

 any fluctuation at all, purple-flowered individuals, when 

 mated entirely at random with white-flowered individuals, 

 should produce progenies consisting of 25, 37.5, 50, 75 or 

 100 per cent, purple-flowered offspring, instead of only 

 50 or 100 per cent. 



I have in several cases found purple-flowered individ- 

 uals among the offspring of two white-flowered parents. 

 Such occurrence was entirely incomprehensible to me ex- 

 cept on the basis of an error in technique. It now be- 

 comes obvious that white crossed with white must oc- 

 casionally give various proportions of purple-flowered 

 offspring. Some such crosses will give all purple while 

 others will give purple and white in ratios 1:1, 3:5, or 

 1 : 3, although a frequent result will be a progeny of all 

 white, which latter alone was expected under the con- 

 ception that the purple color was due to the presence of 

 a single gene. Although I have reared many families in 

 which both parents were white, I have almost invariably 

 obtained a progeny of all whites, but this is undoubted]y 

 due to the fact that such crosses of white with white were 



