As the flowers which were crossed were never castrated, it is probable or even almost certain that I sometimes failed to cross-fertilise them effectually, and that they were afterwards spontaneously self-fertilised. This would have been most likely to occur with dichogamous species, for without much care it is not easy to perceive whether their stigmas are ready to be fertilised when the anthers open. But in all cases, as the flowers were protected from wind, rain, and the access of insects, any pollen placed by me on the stigmatic surface whilst it was immature, would generally have remained there until the stigma was mature; and the flowers would then have been crossed as was intended. Nevertheless, it is highly probable that self-fertilised seedlings have sometimes by this means got included amongst the crossed seedlings. The effect would be, as in the former case, not to exaggerate but to diminish any average superiority of the crossed over the self-fertilised plants.
Errors arising from the two causes just named, and from others,--such as some of the seeds not having been thoroughly ripened, though care was taken to avoid this error--the sickness or unperceived injury of any of the plants,--will have been to a large extent eliminated, in those cases in which many crossed and self-fertilised plants were measured and an average struck. Some of these causes of error will also have been eliminated by the seeds having been allowed to germinate on bare damp sand, and being planted in pairs; for it is not likely that ill-matured and well-matured, or diseased and healthy seeds, would germinate at exactly the same time. The same result will have been gained in the several cases in which only a few of the tallest, finest, and healthiest plants on each side of the pots were measured.
Kolreuter and Gartner have proved that with some plants several, even as many as from fifty to sixty, pollen-grains are necessary for the fertilisation of all the ovules in the ovarium. (1/9. 'Kentniss der Befruchtung' 1844 page 345. Naudin 'Nouvelles Archives du Museum' tome 1 page 27.) Naudin also found in the case of Mirabilis that if only one or two of its very large pollen-grains were placed on the stigma, the plants raised from such seeds were dwarfed. I was therefore careful to give an amply sufficient supply of pollen, and generally covered the stigma with it; but I did not take any special pains to place exactly the same amount on the stigmas of the self-fertilised and crossed flowers. After having acted in this manner during two seasons, I remembered that Gartner thought, though without any direct evidence, that an excess of pollen was perhaps injurious; and it has been proved by Spallanzani, Quatrefages, and Newport, that with various animals an excess of the seminal fluid entirely prevents fertilisation. (1/10. 'Transactions of the Philosophical Society' 1853 pages 253-258.) It was therefore necessary to ascertain whether the fertility of the flowers was affected by applying a rather small and an extremely large quantity of pollen to the stigma. Accordingly a very small mass of pollen-grains was placed on one side of the large stigma in sixty-four flowers of Ipomoea purpurea, and a great mass of pollen over the whole surface of the stigma in sixty-four other flowers. In order to vary the experiment, half the flowers of both lots were on plants produced from self-fertilised seeds, and the other half on plants from crossed seeds. The sixty-four flowers with an excess of pollen yielded sixty-one capsules; and excluding four capsules, each of which contained only a single poor seed, the remainder contained on an average 5.07 seeds per capsule. The sixty-four flowers with only a little pollen placed on one side of the stigma yielded sixty-three capsules, and excluding one from the same cause as before, the remainder contained on an average 5.129 seeds. So that the flowers fertilised with little pollen yielded rather more capsules and seeds than did those fertilised with an excess; but the difference is too slight to be of any significance.