We thus see that the self-fertilised plants which were grown in the nine pots were superior in height (as 116 to 100), and in weight (as 118 to 100), and apparently in hardiness, to the intercrossed plants derived from a cross between the grandchildren of the Brazilian stock.  The superiority is here much more strongly marked than in the second trial with the plants of the English stock, in which the self-fertilised were to the crossed in height as 101 to 100.  It is a far more remarkable fact—­if we bear in mind the effects of crossing plants with pollen from a fresh stock in the cases of Ipomoea, Mimulus, Brassica, and Iberis—­that the self-fertilised plants exceeded in height (as 109 to 100), and in weight (as 118 to 100), the offspring of the Brazilian stock crossed by the English stock; the two stocks having been long subjected to widely different conditions.

If we now turn to the fertility of the three lots of plants we find a very different result.  I may premise that in five out of the nine pots the first plant which flowered was one of the English-crossed; in four of the pots it was a self-fertilised plant; and in not one did an intercrossed plant flower first; so that these latter plants were beaten in this respect, as in so many other ways.  The three closely adjoining rows of plants growing in the open ground flowered profusely, and the flowers were incessantly visited by bees, and certainly thus intercrossed.  The manner in which several plants in the previous experiments continued to be almost sterile as long as they were covered by a net, but set a multitude of capsules immediately that they were uncovered, proves how effectually the bees carry pollen from plant to plant.  My gardener gathered, at three successive times, an equal number of ripe capsules from the plants of the three lots, until he had collected forty-five from each lot.  It is not possible to judge from external appearance whether or not a capsule contains any good seeds; so that I opened all the capsules.  Of the forty-five from the English-crossed plants, four were empty; of those from the intercrossed, five were empty; and of those from the self-fertilised, nine were empty.  The seeds were counted in twenty-one capsules taken by chance out of each lot, and the average number of seeds in the capsules from the English-crossed plants was 67; from the intercrossed, 56; and from the self-fertilised, 48.52.  It therefore follows that:—­

The forty-five capsules (the four empty ones included) from the English-crossed plants contained 2747 seeds.

The forty-five capsules (the five empty ones included) from the intercrossed plants contained 2240 seeds.

The forty-five capsules (the nine empty ones included) from the self-fertilised plants contained 1746.7 seeds.

The reader should remember that these capsules are the product of cross-fertilisation, effected by the bees; and that the difference in the number of the contained seeds must depend on the constitution of the plants;—­that is, on whether they were derived from a cross with a distinct stock, or from a cross between plants of the same stock, or from self-fertilisation.  From the above facts we obtain the following ratios:—­