ii66 A TEXTBOOK OF THEORETICAL BOTANY 



and genera, notably the Primulaceae, the Compositae and the genus Datura 

 in the Solanaceae, all of which are sympetalous. There is a good deal of 

 comparative evidence to support the view that these exceptional marginal 

 bundles, which do not originate as branches from the main petal-traces, are 

 derived from lost stamens of an outer whorl alternating with the petals 

 {i.e., antisepalous), the persistent traces of which have become incorporated 

 into the petal-whorl. Some flowers, indeed, show that their antisepalous 

 stamens have become transformed into supernumerary petaloid structures 

 which are bodily incorporated into the corolla. As all the facts tend to 

 support the idea that the normal petals are sterilized stamens, there is 

 nothing inherently unlikely in the addition to the corolla of these accessory 

 members from the outer stamen-whorl. (See Fig. 1065.) 



The vascular bundles which form the petal supply are usually small and 

 of relatively weak development. Very often they are centric, with phloem 

 surrounding a xylem strand, which, in a few rare instances, as in the 

 corolla of some Gentianaceae, may have a central protoxylem, i.e., the 

 bundle is mesarch, a curious recrudescence of a type of vascular structure 

 long ago lost in the ancestry of the Angiosperms. We shall refer to this 

 again in considering the anatomy of stamens. 



On the much-debated subject of the evolution of the perianth it is only 

 necessary to recapitulate here certain evidence which has been already dealt 

 with in the foregoing pages. The distinction between sepals and involucral 

 bracts is not hard and fast, and in a good many flowers no boundary line 

 can be drawn, as the succession of bracts is uninterrupted right up to the 

 corolla. Several genera, in particular A?iettiotie, also present a variety of 

 conditions among their species, showing the approach towards the flower 

 of a whorl of leafy bracts, from a position well down the pedicel, to a posi- 

 tion immediately below the perianth, when it becomes sophistical to deny 

 it the status of a calyx. 



We have referred to the condition in Mirabilis (see p. 1 143), where only 

 the presence of the lateral flowers of the cyme distinguishes the involucre 

 from a true calyx, and in some species of the genus these flowers are sup- 

 pressed. The morphological character of sepals, their three-trace vascular 

 supply and their histological structure, which is leaf-like, all point in the 

 same direction, namely that they are closely related to bracts and belong 

 to the foliage system. 



Petals, on the contrary, despite their flattened dorsiventral symmetry, 

 seem to be associated with the androecium and to have originated by the 

 sterilization of the lower staminal members. Like stamens they have only 

 a single vascular trace and intermediate forms between petals and stamens 

 are often present. When, for example, members of the normal staminal 

 whorls become partially or wholly sterilized, they assume petaloid forms and, 

 in double flowers, the whole of the stamens may be transformed into extra 

 petals, but not into sepals. The status of petals as sterilized stamens is also 

 supported by their tardy appearance in the ontogeny of the flower bud and 

 by the inverse correlation of the numbers of the two sets of organs which is 



