360 HYBRID OF DIGITALIS. 



i ■ /Breadth inr\r\ 



in the two species; the ratio I ^ -7— x 10001 being 684 m 



gloxiniazflora, 685 in lutea, and 786 in hybrid large seeds and 510 

 in hybrid small seeds. Both the large and small seeds were com- 

 monly capable of germination; but in the case of the small, narrow 

 seeds the cotyledons had the greatest difficulty in freeing them- 

 selves from the testa. All the hybrid plants which were subse- 

 quently raised were almost certainly derived from the large seeds. 



In the development of a normal foxglove seed the nucellus 

 forms a very regular tapetal layer of columnar cells around the 

 embryo-sac (Plate V, fig. 1). As the sac grows, this layer, with 

 the disappearance of the remainder of the nucellus, becomes pushed 

 against the testa, and in the ripe seed the cell-contents disappear 

 entirely, and the cell-walls cease to exhibit cellulose reactions, as 

 shown by being insoluble in strong acids, staining intensely with 

 Sudan III, and by a number of other tests. In ripe, normal seeds 

 the tapetal layer is one cell thick, and this is also generally the 

 case in the large, hybrid seeds (fig. 6); but in the small seeds the 

 layer may be 4 or 5 or more cells thick (fig. 5, t. I. e. c), the con- 

 dition arising by the more or less regular tangential division of the 

 original layer of cells enveloping the embryo-sac. The hypertrophy 

 of this layer is probably to be regarded as a teratological condition 

 arising through the weak development of the embryo and 

 endosperm. In sections through hybrid seeds, which were found 

 to be incapable of germination, it was seen that sometimes both 

 embryo and endosperm were greatly reduced (figs. 3, 4), or com- 

 pletely absent, and nearly the whole bulk of the original nucellus 

 area was occupied by a thick mass of empty, small, thin-walled cells 

 arranged in radial rows and derived from the excessive tangential 

 division of the single layer of tapetal cells. 



2. LUTEA (Q) X GLOXINICEFLORA ((f). 



The reciprocal cross was tried, with lutea as the seed-plant. 

 Ten flowers in each of four lutea plants were fertilised with the 

 pollen of gloxinia flora, 25 plants being used. Out of the 40 

 ovaries some 29 showed signs of development and were harvested. 

 On examination it was found that 18 contained nothing but dried 

 ovules, 5 contained only shrivelled seed, and 6 bore one or more 

 seeds of normal appearance and many shrivelled seeds. The 

 average number of such normal seeds in the 6 capsules was only 

 two. Of these 12 seeds only one germinated with any vigour, and 

 the very young seedling subsequently ceased to grow and died. It 

 is thusi quite clear that with lutea as the seed-plant the cross is 

 much less easily effected than when gloxiniceflora is the seed-plant. 

 The hybrid seeds did not exhibit a similar dimorphic condition as 

 seen when gloxiniceflora was the seed-plant; they were either more 

 or less normal in appearance, or were shrivelled and totally incap- 

 able of germination. 



