264 Shull. 



plants in Family 0745, coming as they did from seeds of the hep- 

 iandra-iorra, should therefore prove to be heterozygous. This ex- 

 pectation was fully realized, as will be noted in the following table 

 which gives the results of crosses among these normal-flowered in- 

 dividuals, and between them and their abnormal sibs. 



Family 



0930 

 0932 



0931 

 0936 



0934 

 0937 



0935 

 "933 



Only in crosses between abnormal white and normal purple 

 (Families 0934 and 0937) is there a departure from expectation suffi- 

 ciently great to suggest that the discrepancies are possibly not solely 

 due to errors of random sampling. In the offspring of these reciprocal 

 crosses the number of normal purple individuals is nearly 40 per 

 cent in excess of expectation. Considering the small numbers em- 

 ployed, this departure should not be given undue weight, however, 

 especially as the deviation comes within three times the probable 

 error, which is usually taken as the limit beyond which a variation 

 may be pretty safely assumed to be due to some other than a chance 

 distribution. Accepting this view of the exceptionally large proportion 

 of normal purple-flowered individuals in this one pair of reciprocal 

 crosses, it is seen that the constitution of all these families is in 

 accord with the view that the lieptandra-ioxr^ differs from the normal 

 Digitalis purpurea in a single Mendelian unit-character, and that it 

 is recessive to the normal form. 



While the origin of such new forms by gradual changes in the 

 genotype is not absolutely unthinkable, the recent accumulations of 



1) N = Normal; A = Abnormal; P = Purple; W = White. 



