Deducting these fourteen cases, there remain twelve in which the crossed plants show no well-marked advantage over the self-fertilised. On the other hand, we have seen that there are fifty-seven cases in which the crossed plants exceed the self-fertilised in height by at least five per cent, and generally in a much higher degree. But even in the twelve cases just referred to, the want of any advantage on the crossed side is far from certain: with Thunbergia the parent-plants were in an odd semi-sterile condition, and the offspring grew very unequally; with Hibiscus and Apium much too few plants were raised for the measurements to be trusted, and the cross-fertilised flowers of Hibiscus produced rather more seed than did the self-fertilised; with Vandellia the crossed plants were a little taller and heavier than the self-fertilised, but as they were less fertile the case must be left doubtful. Lastly, with Pisum, Primula, the three generations of Canna, and the three of Nicotiana (which together complete the twelve cases), a cross between two plants certainly did no good or very little good to the offspring; but we have reason to believe that this is the result of these plants having been self-fertilised and cultivated under nearly uniform conditions for several generations. The same result followed with the experimental plants of Ipomoea and Mimulus, and to a certain extent with some other species, which had been intentionally treated by me in this manner; yet we know that these species in their normal condition profit greatly by being intercrossed. There is, therefore, not a single case in Table 7/A which affords decisive evidence against the rule that a cross between plants, the progenitors of which have been subjected to somewhat diversified conditions, is beneficial to the offspring. This is a surprising conclusion, for from the analogy of domesticated animals it could not have been anticipated, that the good effects of crossing or the evil effects of self-fertilisation would have been perceptible until the plants had been thus treated for several generations. The results given in Table 7/A may be looked at under another point of view. Hitherto each generation has been considered as a separate case, of which there are eighty-three; and this no doubt is the more correct method of comparing the crossed and self-fertilised plants.