372 HYBRID OF DIGITALIS. 



strict Mendelian ; but it follows naturally on the conception of 

 factors capable of being gradually modified by the influence of other 

 factors and perhaps also by the external environment. 



In the case of the peloric condition of foxgloves, starting with 

 homozygous plants in which this character was quite definitely 

 present, it has been possible by selection of self-fertilised genera- 

 tions to eliminate the character altogether. A similar result was 

 reached, but with much greater difficulty, in obtaining a white 

 foxglove by the selection of self -fertilised generations of homozy- 

 gous, pale mauve plants. Quantity of a character being inherited 

 by the blending type, and selection being effective, it is sometimes- 

 possible to eliminate a character altogether, and the point of 

 contact between Mendelian and blending inheritance is reached. 



These results indicate a gradual change in the nature of the 

 factor or factors governing a blending character, and the change 

 can arise through crossing, the homologous factors mutually 

 influencing each other. 



In the present paper distinct species are dealt with having 

 very marked specific differences. The character in the interspecific 

 hybrid may be intermediate or considerably nearer to one species 

 or the other, and thus dominance or prepotency may be exhibited. 

 Similarly in the backcrosses the character may be intermediate, 

 or much nearer either to the hybrid parent or to the pure species 

 parent. 



On account of sterility, raising offspring by self-fertilisation 

 of the hybrid was not possible ; but for testing the theory of 

 segregation and the persistence of the purity of factors, backcrosses 

 are equally useful. In all the measurable characters examined 

 there was no clear reappearance in the backcrosses of the grand - 

 parental characters in an uncontaminated condition. The view 

 that these results are due in every case to the existence of multiple, 

 independent, unchangeable factors would be only justified if the 

 backcross offspring exhibited exceptional variability, but such is 

 not the case. Thus the results obtained with interspecific hybrids 

 resemble those found with intervarietal hybrids. 



The conception of the segregation of factors in the division 

 of cells is useful, provided it is not demanded by the Mendelians 

 that these factors shall necessarily have any fixed or invariable 

 nature. In most ordinary, measurable characters the factors on 

 crossing would seem to blend, and although the blend may not 

 necessarily entail an intermediate condition of the character, yet 

 the factors, or connected group of factors, are changed by the 

 crossing, and a reappearance of the pure character is not to be 

 expected, and would only result by the production of a new factor, 

 and not by the persistence of the original, unchanged one. 



Explanation of Plates. 



Plate IV. 



Fig. 1. — GIox. pollinated with In ten yielded dimorphic seeds 

 (large to small about 1:75 in number) both capable of germina- 



