We see in this table that the crossed plants are always in some degree more productive than the self-fertilised plants, by whatever standard they are compared.  The degree differs greatly; but this depends chiefly on whether an average was taken of the seeds alone, or of the capsules alone, or of both combined.  The relative superiority of the crossed plants is chiefly due to their producing a much greater number of capsules, and not to each capsule containing a larger average number of seeds.  For instance, in the third generation the crossed and self-fertilised plants produced capsules in the ratio of 100 to 38, whilst the seeds in the capsules on the crossed plants were to those on the self-fertilised plants only as 100 to 94.  In the eighth generation the capsules on two self-fertilised plants (not included in table 2/18), grown in separate pots and thus not subjected to any competition, yielded the large average of 5.1 seeds.  The smaller number of capsules produced by the self-fertilised plants may be in part, but not altogether, attributed to their lessened size or height; this being chiefly due to their lessened constitutional vigour, so that they were not able to compete with the crossed plants growing in the same pots.  The seeds produced by the crossed flowers on the crossed plants were not always heavier than the self-fertilised seeds on the self-fertilised plants.  The lighter seeds, whether produced from crossed or self-fertilised flowers, generally germinated before the heavier seeds.  I may add that the crossed plants, with very few exceptions, flowered before their self-fertilised opponents, as might have been expected from their greater height and vigour.

The impaired fertility of the self-fertilised plants was shown in another way, namely, by their anthers being smaller than those in the flowers on the crossed plants.  This was first observed in the seventh generation, but may have occurred earlier.  Several anthers from flowers on the crossed and self-fertilised plants of the eighth generation were compared under the microscope; and those from the former were generally longer and plainly broader than the anthers of the self-fertilised plants.  The quantity of pollen contained in one of the latter was, as far as could be judged by the eye, about half of that contained in one from a crossed plant.  The impaired fertility of the self-fertilised plants of the eighth generation was also shown in another manner, which may often be observed in hybrids—­namely, by the first-formed flowers being sterile.  For instance, the fifteen first flowers on a self-fertilised plant of one of the later generations were carefully fertilised with their own pollen, and eight of them dropped off; at the same time fifteen flowers on a crossed plant growing in the same pot were self-fertilised, and only one dropped off.  On two other crossed plants of the same generation, several of the earliest flowers were observed to fertilise themselves and to produce capsules.  In the plants of the ninth, and I believe of some previous generations, very many of the flowers, as already stated, were slightly monstrous; and this probably was connected with their lessened fertility.