most frequently broken, and especially in those flowers having an indefinite number of stamens and petals. In such plants in the wild state there is usually no established uniform number for eit her of these parts, and it may be that they vary as circumstances determine. In such cases it seems more natural to suppose that one gives place to the other, than that there is an independent development of a new part. However, when we come to the cultivated plants, this seeming chasm between petals and stamens is bridged, and the difficulty now turns upon deciding whether a certain organ is more or less stamen than petal.
As seen from both a physiological and morphological point of view, the pistil is considered the most highly differentiated part of the flower, the stamen next, petal next, and sepal least. Under the conditions obtaining with the cultivated rose, stamens are less important than petals, and probably less easily produced. Instead of the slender filament surmounted by the two lobes of the anther, bearing thousands of expensive pollen-grains, there is a broad, loose-celled, showy petal. When a stamen is replaced by a petal, it is naturally termed retrograde metamorphosis. In the rose, as in many other cultivated plants, all gradations may be found, from a normal stamen, with a slight color-line along one side of the slender filament, to the perfect petal, which may have a small notch at the tip, marking where the anther might have been. So long as the demand for self-propagation is met in other ways, the tendency to produce seed may be overcome, and the plant spends its energies in the formation of showy blossoms, possibly losing, for the time at least, the power to ripen seed. If the selective power of the rosarian is now withdrawn, while at the same time the stimulation of high culture remains, the inference is not unwarranted that the retrogression would continue so far that no flowers develop. It may be that the so-called green roses sometimes met with furnish solid ground for such a view. At any rate, they are forcible examples of the throwing off of floral disguises, and the true nature of the parts becomes evident to the most skeptical.
Rosaceous flowers furnish examples of the simplest form of doubling. In many others the struggle between the two forces seems to have been more violent, and the results are far from uniform, even in the same flower. In some species the broadening of the stamen takes place above the anther, as if the filament had become prolonged and petaloid. Frequently with such structures the rudimentary anther is at the base of the petal, or one half is midway upon one side, and the other opposite it, the connective having broadened out into the body of the petal. It is not unusual to find one half of the organ petaloid, while the other is contracted, contorted, and bears an anther-lobe containing fully