THE ANGIOSPERMAE 



1119 



with the stamens. The sepals would be therefore interpreted as modified 

 bracts, not as sterilized " sporophylls ", while the petals would be assumed 

 to belong to the potentially fertile portion of the flower system. 



The stamens have become consistently i -trace organs in most flowers, 

 but in a few primitive families, especially Magnoliaceae and Winteraceae, 

 3-trace stamens exist, so that presumably all the floral parts originally 

 conformed to the 3-trace pattern, which Eames has shown good reason 

 for supposing to have been the foundational pattern in the leaves of the 

 vegetative axis. This is not of course to be taken as implying that the floral 

 parts are derived directly from foliar leaves or vice versa, an outworn con- 

 ception, but simply that they have a common morphological character. 



Fig. 1095. — Diagrams of theoretical primitive carpels, showing palmate vena- 

 tion and 3, 5 and 7 traces from as many gaps. (After Eames.) 



The carpels share the common 3-trace structure, with variations of i, 

 3 and 5 to many traces (Fig. 1095). The 3-trace condition is the commonest 

 and is above all characteristic of follicles. The i -trace carpels are nearly 

 always akenes, the ventral bundles, which supply the ovules, branching 

 off from the single trace, which is dorsal in the carpel. The ventral traces 

 usually persist somewhat above the insertion of the ovule, if this is near the 

 base of the akene, and in a few cases the upper parts of these traces produce 

 branches either to abortive ovule rudiments, e.g., in Clematis recta, or even 

 to the places where these might be expected, but from which they are 

 absent, e.g., Anemone. This is credible evidence that the akene condition 

 has been reached by reduction from multiovulate carpels of the follicle 

 type. It is also evidence for the important anatomical consideration that the 



