366 



HYBRID OF DIGITALIS. 



By backcrossing a hybrid plant with both of the parent species r 

 and examining a series of widely different characters in the off- 

 spring, it is considered that some evidence is obtainable as to the 

 condition of the factors in the hybrid parent; that is to say, 

 whether the two factors of an homologous pair have retained their 

 differences unimpaired, so that on segregation occurring the 

 characters would reappear in their original purity, or whether 

 they have mutually influenced each other, or blended, to form a 

 pair of similar factors which may or may not be intermediate iiL 

 nature between the two original factors. 



In order to see the bearing of the experiment on the problem, 

 it is insufficient to deal simply with means ; it is necessary to give 

 the frequency distribution of variation in the form of tables which 

 will show at a glance the range of variability in the different 

 generations. In this place we will deal with a few of the charac- 

 ters which have been investigated. 



Flower: Length of Corolla.— The length of the flower of 

 lutea is very much less than in gloxinia 'flora •. In the hybrid the 

 length tends to be nearer to that of gloximceflora (PI. IV, fig. 2). 

 The result of backcrossing the hybrid with the two parent species 

 is shown in the accompanying table. The mean of the gloxinia flora 

 backcross (500) is much nearer to the gloxinia flora mean (505) 

 than to the hybrid mean (421), and the mean of lutea backcross 

 (356) is nearer to the lutea mean (318) than to the hybrid mean 

 (421). 



Absolute Length of Corolla. 



Supposing that corolla-length is controlled by one factor, or 

 a group of connected factors, the gametic constitution, according 

 to Mendelian theory, is given at the head of the columns. 



