116 



JOURNAL OF THE ROYAL HORTICULTURAL SOCIETY. 



Here, therefore, we fxem to see some approximation to a carpel, such 

 as a pea-pod when split open down the coherent margins. The pod then 

 resembles this carpellary leaf-scale of a Cycad with its two rows of ovules. 



There is yet a third family of Conifers, comprising three genera only, 

 which supplies us with something like a perianth, and the appearance, at 

 least, of an ordinary flower. 



Gnctaccce, like the Conferee, have the sexes separate, and each is pro- 



FiG. 36. — Carpellavy scale of Cypress, Fig, 37. — Carpellary leaf of C?/cas, with 



with many erect ovules. six ovules on the margins. 



tected by a perianth, consisting of four separate leaves or bract-like 

 structures in Wchcltschia, and a gamophyllous one in Gnctum and 

 Ephedra ; but there is no trace of a pistil. 



Besides the female flower of Welwitscliia there is a male flower, but 

 hermaphrodite in form, as it contains a central ovule, but it is abortive 

 (fig. 38). It consists of a perianth composed of two pairs of scale-like 

 leaves, several stamens in a coherent whorl, and an ovule in the centre. 

 This latter, however, though it has a styliform process above, is closed at 



a b 



Fig. 38. — Male flower of TUcZm^sc/iia. Fig. 39. -a, Male flower ; b, female 



flower of Sallow Willow. 



the apex ; so that the flower is really male only. As Wclicitschia has 

 never more than one pair of leaves besides the cotyledons, we can trace 

 the source of the two pairs of decussate leaves of the perianth from the 

 original opposite conditions of its primordial leaves. The scales of the 

 cones .of this plant are similarly arranged. 



The next step is to see if we can find plants among angiospermous 

 Dicotyledons which point to any affinity with Gymnosperms. Herein lies 



