That these flowers owe their structure primarily to the arrested development of perfect ones, we may infer from such cases as that of the lower rudimentary petal in Viola being larger than the others, like the lower lip of the perfect flower,--from a vestige of a spur in the cleistogamic flowers of Impatiens,-- from the ten stamens of Ononis being united into a tube,--and other such structures. The same inference may be drawn from the occurrence, in some instances, on the same plant of a series of gradations between the cleistogamic and perfect flowers. But that the former owe their origin wholly to arrested development is by no means the case; for various parts have been specially modified, so as to aid in the self-fertilisation of the flowers, and as a protection to the pollen; for instance, the hook-shaped pistil in Viola and in some other genera, by which the stigma is brought close to the fertile anthers,- -the rudimentary corolla of Specularia modified into a perfectly closed tympanum, and the sheath of Monochoria modified into a closed sack,--the excessively thin coats of the pollen-grains,--the anthers not being all equally aborted, and other such cases. Moreover Mr. Bennett has shown that the buds of the cleistogamic and perfect flowers of Impatiens differ at a very early period of growth. The degree to which many of the most important organs in these degraded flowers have been reduced or even wholly obliterated, is one of their most remarkable peculiarities, reminding us of many parasitic animals. In some cases only a single anther is left, and this contains but few pollen-grains of diminished size; in other cases the stigma has disappeared, leaving a simple open passage into the ovarium. It is also interesting to note the complete loss of trifling points in the structure or functions of certain parts, which though of service to the perfect flowers, are of none to the cleistogamic; for instance the collecting hairs on the pistil of Specularia, the glands on the calyx of the Malpighiaceae, the nectar-secreting appendages to the lower stamens of Viola, the secretion of nectar by other parts, the emission of a sweet odour, and apparently the elasticity of the valves in the buried capsules of Viola odorata. We here see, as throughout nature, that as soon as any part or character becomes superfluous it tends sooner or later to disappear.