In the accompanying figures (fig. 163, _a-d_) the nature of this change is illustrated. In some of the specimens it is easy to see that the two shorter stamens undergo the change into carpels later and less perfectly than the four longer ones, and not infrequently the outer pair are altogether absent. In most of the flowers of this variety the petals are smaller and less perfectly developed than usual.[341] [Illustration: FIG. 163.--_Cheiranthus Cheiri_, var. _gynantherus_. _a._ Sepals and petals removed to show carpellodic stamens. _b._ The same laid open. _c._ Transverse section. _d._ Plan of flower with four carpel-like stamens, &c.] In _Lilium tigrinum_, some specimens of which were gathered by Mr. J. Salter, in addition to various degrees of synanthy and other changes, some of the stamens were developed in the form of carpels, adherent by their edges so as to form an imperfect tube or sheath around the normal pistil. Fig. 164 shows one of the intermediate organs from these flowers, in which half the structure seems devoted to the formation of ovules, while the other half bears a one-celled anther. Lindley[342] has also described a case of this kind in a species of _Amaryllis_. [Illustration: FIG. 164.--Structure half anther, half carpel, _Lilium_.] In _Saxifraga crassifolia_ it sometimes happens that mixed with the stamens, and originating with them, are a number of distinct and perfectly formed carpels, wholly separated from the normal carpels, in the centre of the flower. In this particular instance there is usually no intermediate condition between the stamen and the pistil. Guillemin[343] also describes a transformation of the stamens into carpels in _Euphorbia esula_. When the anther is involved it may be only partially so, or almost the whole organ may be transformed. As instances of very partial change may be cited the passage of the connective into a stigma in _Thalictrum minus_, or the passage of the points of the anthers into imperfect styles in some species of bamboo.[344] In _Rosa arvensis_ similar transformations have been observed of a slightly more complex character than those just mentioned, and passing into more important changes, especially to the formation of pollen within ovules, formed on the edges of an open carpellodic anther (see p. 186). Mr. Berkeley has recorded an analogous case in a gourd in which the stamens bore numerous ovules (p. 200), and Baillon describes another gourd in which certain fleshy appendages surrounding the androecium were provided with ovules.[345] Payer, in his 'Organogenie,' p. 38, mentions a stamen of _Dionaea_ bearing not only an anther, but likewise an ovule.