288 Transactions of the Society. 



This would make it lighter and enable it to float, or perhaps 

 prevent it from sinking too deeply into the herbage of the Sphagnum 

 in which it so often lives. 



Each of our species of Drosera differs somewhat from the others 

 in the texture of the surface of the seeds. There must, I suppose, 

 be some reasons for these differences, but they are not very 

 apparent. 



In D. intermedia the seeds are densely covered with small 

 elevated points, as in some species of Arenaria, Silene, and other 

 Caryophylleae. These would, no doubt, lighten the seeds. 



Umbellifeile. — There are two carpels, coherent into a 2-celled 

 ovary, each cell containing one ovule, suspended from the top. 

 The fruit is 2-celled, dividing into two portions (mericarps) often 

 suspended at the top of single or double axis. The surface has 

 ten ridges, sometimes produced into wings. The furrows between 

 the main elevations are sometimes occupied by subordinate ridges. 

 The seed is pendulous. The fruit is often compressed ; sometimes 

 laterally, in which case a slice cut through the seed has an oval 

 form, the division being across the narrow diameter. When the 

 compression is from back to front, the division is across the 

 broadest diameter. In this order the seeds are comparatively 

 uniform, and the main differences occur in the fruits. 



The fruits are dry and in some cases eaten by birds, but the 

 principal modes of dispersal are by hooks or wings. 



It might have been expected that these different methods of 

 dispersal would have prevailed in different groups of the order. 

 As, however, we have seen in other cases, this is not the case. 

 Hooks, for instance, occur in several genera (Sanicula, Anthriscus, 

 JDaucus, Caucalis) by no means nearly allied. Anthriscus vulgaris, 

 for instance, in which the carpels are armed with hooked bristles, 

 is so nearly allied to Chcerophyllum temulum and G. sylvestre that 

 Bentham in the " Handbook of the British Flora " places it in the 

 same genus as C. Anthriscus. 



In Eryngium the carpels are covered with chaffy scales, which 

 are longest on the primary ridges. These would serve to lighten 

 the fruit, but they would also help to entangle them in the fur of 

 animals. 



In other cases the persistent styles are recurved, forming 

 hooks which would serve for the same purpose [uEgoptodium, Slum, 

 Pimpinclla). 



Winged fruits occur in Angelica, Smyrnium, Crithmum, 

 Myrrhis, Sium, etc. These, as we should expect, are glabrous. 



In Scandix Pecten-veneris the fruit is developed into a long 

 beak, and when the bases of the carpels split away, they diverge 

 widely. Perhaps this facilitates their being torn off by any passing 

 animals. 



The aquatic species, as usual, are glabrous. In some Umbellifers, 



