MR. J. G. BAKEk's SYSTEMA IllIDACEAllUM. 09 



i 



except that in the monotypic genus Jlermodactylus the three 

 placentas are parietal, and t!ie ovary conseciuently one-celled. 

 The ovules in each cell are always more than one and super- 

 posed. In style and stigma there is a very wide variation 

 within the bounds of the order. We can always trace out three 

 distinct style-branches. Sometimes, as in most of the Euiridea?, 

 these are opposite to the stamens, and consequently to the three 

 outer segments of the perianth ; but in another large section of 

 the order, as in Sisyrinchium, Lihertia, and Patersonia^ the style-, 

 branches are alternate with the stamens, and consequently op- 

 posite to the three inner segments of the perianth. In Aristea the 

 three branches are mere cusps ; but, as a rule, they are of con- 

 siderable length. Sometimes they are subulate, and then may 

 be either entire or more or less deeply forked. A common type 

 is that of Gladiolus, in which the stigmas are represented hj small, 

 flat, oblauceolate, petaloid expansions. We have an extreme 

 type in the large, petaloid, crested style-branches of Iris and 

 Morcea, As a whole, we may safely say that some of the best 

 characters of genera are taken from these organs ; and, as a 

 rule, it seems that we may safely hold a material difference in 

 style as the mark of a distinct genus, although in Crocus we get 

 a wide range of diversity in plants otherwise very near to one 



another. 



Fruit and Seeds, — The fruit is always capsular, with loculicidal 

 dehiscence. In some cases, as for instance, Tigridia^ Migidella^ 

 and Herbert ia, the top is protruded beyond the perianth- tube, 

 and this leaves a distinct free operculum at the top of the cap- 

 sule, at the bottom of which, at any rate temporarily, the dehis- 

 cence stops. The seeds are generally angular; and in those 

 genera in w hich they are typically discoid we get species in which 

 the discoid character is lost. Gladiolus segetum^ wath globose 

 seeds, is very difficult to distinguish from G. hyzantinus^ in which 

 they are as flat as in a tulip. Iris and Antholyza are genera in 

 which we get a very individualized type of flower in combination 

 with seeds of a very different shape. So that for purposes of sys- 

 tematic distinction the seeds in the order are not readily avail- 

 able. The testa in the order differs materially in colour and tex- 

 ture, and needs greater study of fresh specimens than I have been 

 able to giA^e. In embryo and albumen I know of no material 

 difference within the bounds of the order. 



