HYBRID OF DIGITALIS. 363 



bility in size, but there was no sharp dimorphism into large and 

 small seeds, as seen in capsules resulting from the cross of 

 gloxinimflora (Q) with lutea (d*). 



On this hybrid plant, 20 flowers were pollinated from lutea, 

 and capsules developed ; 3 of these capsules contained no good seed, 

 while each of the remaining 17 bore on the average 14 good seed? 

 of small size. On the same hybrid plant, 30 flowers were crossed 

 with the pollen from several (/lux ink? flora plants. Of the resulting 

 capsules 9 contained no good seeds, while each of the remaining 21 

 bore on the average only 4 good seeds. Thus the cross, hybrid 

 x lutea, was considerably more fertile than the cross, hybrid x 

 gloxinia; flora. Bv "good" seed is meant unshrivelled seeds having 

 a proper shape and stoutness, it does not necessarily mean that thev 

 were all capable of germination. 



5. The Seeds of the Hybrid, obtained with glox, and lutea 



Pollen. 



In identically the same hybrid plant the seeds obtained bv 

 using gloxi nice flora pollen differed distinctly in external appear- 

 ance from those formed when lutea pollen was employed. 



a. The testa was of a dark chestnut brown in seeds- 



obtained with gloxiniasflora pollen, and of a yellow 

 colour with lutea pollen. The distinction was not 

 due to any difference in the character of the 

 endosperm, for this was white in both. 



b. The mean ratio, -= r— x 1000, of 81 seeds (from 



Length 



10 capsules in all) with gloxiniasflora pollen was 628, 

 and of 75 seeds (from 9 capsules in all) with lutea 

 pollen was 578. This constitutes a well-marked dif- 

 ference readily detected by the eye (PI. IV, fig. 1). 



c. The cells of the testa were somewhat larger and had 



thicker walls with gloxinia; flora pollen than with 

 lutea pollen, the difference in tangential diameter 

 being about 9:8. The outer tangential wall of the 

 testa cells usually disappeared entirely in seeds from 

 the gloxinia 'flora pollen, but a thin glistening lamina 

 generally persisted in the seeds from lutea pollen 

 (PI. V, fig. 6, p. o. «?.). 



d. The hilum region of the seed projected on the average 



about 4 times more prominently beyond the general 

 contour in seeds from lutea pollen than in those from 

 gloxinia; flora pollen (PI. IV, fig. 1). 



In both kinds of seeds the dead tapetal layer immediately 

 within the testa was of variable thickness, but was always more 

 conspicuously developed than in the larger of the two sizes of seeds 

 derived from the cross, gloxiniceflora and lutea. The general con- 

 dition resembled that shown in PI. V, fig. 5. 



6a 



