This dark purple variety did not appear, as far as my gardener and myself could recollect, before the fifth or sixth self-fertilised generation. However this may have been, it became, through continued self-fertilisation and the cultivation of the plants under uniform conditions, perfectly constant, to the exclusion of every other variety.
Dianthus caryophyllus.
The self-fertilised plants of the third generation all bore flowers of exactly the same pale rose-colour; and in this respect they differed quite remarkably from the plants growing in a large bed close by and raised from seeds purchased from the same nursery garden. In this case it is not improbable that some of the parent-plants which were first self-fertilised may have borne flowers thus coloured; but as several plants were self-fertilised in the first generation, it is extremely improbable that all bore flowers of exactly the same tint as those of the self-fertilised plants of the third generation. The intercrossed plants of the third generation likewise produced flowers almost, though not quite so uniform in tint as those of the self-fertilised plants.
Petunia violacea.
In this case I happened to record in my notes that the flowers on the parent-plant which was first self-fertilised were of a "dingy purple colour." In the fifth self-fertilised generation, every one of the twenty-one self-fertilised plants growing in pots, and all the many plants in a long row out of doors, produced flowers of absolutely the same tint, namely, of a dull, rather peculiar and ugly flesh colour; therefore, considerably unlike those on the parent-plant. I believe that this change of colour supervened quite gradually; but I kept no record, as the point did not interest me until I was struck with the uniform tint of the flowers on the self-fertilised plants of the fifth generation. The flowers on the intercrossed plants of the corresponding generation were mostly of the same dull flesh colour, but not nearly so uniform as those on the self-fertilised plants, some few being very pale, almost white. The self-fertilised plants which grew in a long row in the open ground were also remarkable for their uniformity in height, as were the intercrossed plants in a less degree, both lots being compared with a large number of plants raised at the same time under similar conditions from the self-fertilised plants of the fourth generation crossed by a fresh stock. I regret that I did not attend to the uniformity in height of the self-fertilised seedlings in the later generations of the other species.
These few cases seem to me to possess much interest. We learn from them that new and slight shades of colour may be quickly and firmly fixed, independently of any selection, if the conditions are kept as nearly uniform as is possible, and no intercrossing be permitted. With Mimulus, not only a grotesque style of colouring, but a larger corolla and increased height of the whole plant were thus fixed; whereas with most plants which have been long cultivated for the flower-garden, no character is more variable than that of colour, excepting perhaps that of height. From the consideration of these cases we may infer that the variability of cultivated plants in the above respects is due, firstly, to their being subjected to somewhat diversified conditions, and, secondly, to their being often intercrossed, as would follow from the free access of insects. I do not see how this inference can be avoided, as when the above plants were cultivated for several generations under closely similar conditions, and were intercrossed in each generation, the colour of their flowers tended in some degree to change and to become uniform. When no intercrossing with other plants of the same stock was allowed,--that is, when the flowers were fertilised with their own pollen in each generation--their colour in the later generations became as uniform as that of plants growing in a state of nature, accompanied at least in one instance by much uniformity in the height of the plants.