MINUTE STRUCTURE OF PLANT HYBRIDS. 277 



to the pollen grains, and gave to each a certain and recognizable tint. Now we know that 

 repeated attempts to cross C. purpureus and C. Laburnum have failed, and it is quite 

 possible that the action of the abundant tannin material on the stigma or egg cell of C. 

 Laburnum, and vice versa the small amount of it for C. purpureus, may largely explain 

 the failures. Many of the negative hybridization experiments of the past therefore may 

 have less depth of significance than one is inclined on first view to attach to them. 



(f) On the Relative Fertility of Hybrids in Relation to Heredity. — The concensus of 

 opinion among the older hybridizers was that very few hybrids were fertile, and that 

 those which were, gradually returned to one of the parent types. During the last 

 twenty or twenty-five years the opinion has been freely criticised, and rightly so, since 

 horticulturists in that period have carried forward experimental hybridization by leaps 

 and bounds, and have imported through intelligent collectors not a few wild plants which 

 they regarded as, and in some instances have proved to be, natural hybrids. 



But even though the subject of pollination, with all the marvellous floral adaptations 

 for it, were unknown to such experimenters as Kolreuter, Knight, Gartner, Wichura, 

 and others, the main outcome of their researches can scarcely be set aside, though we 

 may have to give a more liberal interpretation to it in the future. To sum up present- 

 day experiences, it may be said that crosses between species that are nearly related in 

 structure and habit can readily be effected, and the offspring may be largety fertile, at 

 least among certain genera. Crosses between species that differ considerably in form, 

 flower-colour, and habit are more difficult to perform, and the hybrids are largely sterile, 

 while crosses between such divergent species or genera as Dianthus alpinus and 

 barbatus, Saxifraga Geum and Aizoon, Lapageria and Philesia are almost or wholly 

 sterile. 



Now when the pollen and egg cells from each of these three roughly classified groups 

 are examined one finds that a few of those from the first are shrivelled-looking and badly 

 formed ; from the second a considerable percentage are thus affected ; while from the 

 third it is difficult to get one good pollen grain, and rather difficult to get one well-formed 

 egg-cell, though these do not appear to be so much affected as the pollen grains. 



We have repeatedly referred to, and in Plate V. fig. 10 6 have illustrated a bad pollen 

 sample. The cells are always smaller, often greatly smaller, than in either parent ; the 

 protoplasm is devoid of rich nutritive granules and is scanty in quantity, so that it does 

 not fill the cell cavity, the wall is irregular in outline and imperfectly formed. Shortly, 

 therefore, it may be said that while the vegetative cells of a hybrid can develop gradually 

 into organs that are a blended reproduction of those of the parents, the generative cells 

 fail to receive or to form appropriate protoplasmic material. 



Consideration of this causes us to look at the theories that have been advanced to 

 account for heredity. Darwin's theory of pangenesis has been put aside as cumbrous 

 and difficult to conceive of in practice, though it explained phenomena of heredity all 

 along the line better than any previously existing view. More applicable, however, docs 

 Nageli's idioplasmatic theory appear, in spite of gratuitous assumptions that have been 



