36 Mr. J. Miers on the Winteracese. 



but some incompatible features are seen in the monoecious cha- 

 racter of the latter order, and the trimerous arrangement of the 

 parts : we must, however, remember that in Tasmannia there is 

 an approximation in this respect ; for if its flowers be not mon- 

 oecious, they are often polygamous ; and although Drimys gene- 

 rally does not agree with Kadsura and Sphcerostemma in its parts 

 being always in threes or multiples of three, we find these num- 

 bers in Illicium, and sometimes in Drimys; while, on the other 

 hand, Schizandra differs from its congeners in having a penta- 

 merous disposition. There is also in the Winteracece a tendency 

 towards the Schizandracece ; for though Tasmannia has only one 

 or two, or sometimes four ovaries, Sin&Drimys six to eight, Illicium 

 has at times as many as eighteen distinct carpels, — but then, it 

 must be remembered, they are always uniserial. There is also 

 much similarity in the structure of their ovaries, that is to say, 

 they are unilocular, with few ovules attached parietally to the 

 ventral suture. The fruit in both orders is baccate and few- 

 seeded, the seeds being reniform, somewhat small, with a crus- 

 taceous shining external tunic, and a short internal raphe along 

 the sinus, together with an extremely small embryo imbedded 

 at the extremity of the albumen near the hilum, having very 

 minute cotyledons in proportion to the radicle. The Schizan- 

 dracece, like the Winteracece, have alternate exstipulate leaves, 

 which are equally distinguished by numerous pellucid dots ; and 

 I have before alluded to the peculiar dotted vessels contained in 

 the wood of all these three families. In Schizandra, the stamens 

 are united into a dilated tube or disk, and Canella offers some 

 analogy with this; but the principal line of distinction that 

 marks the group of the Schizandracece lies in its monoecious 

 flowers, and in the multiserial disposition of its distinct and 

 numerous carpels. Notwithstanding the well-marked characters 

 that tend to separate these orders, enough has been shown to 

 prove their near affinity. 



The Schizandracece by some botanists are held to be merely a 

 tribe of the Magnoliacece ; but I have pointed out many of the 

 characters that keep them distinct from the latter, and that place 

 them in contiguity with the Winteracece. It is not possible, in 

 any linear arrangement, to form so perfect an approximation of 

 genera as by the circular system ; but the nearest conformity is 

 obtained by interposing the Menispermacece and Lardizabalacece 

 between the Anonacece and Magnoliacece, and by the intervention 

 of Schizandracece between the last-mentioned family and the 

 Winteracece and Canellacece ; by this disposition we form a chain 

 of regular gradation, the links of which sufficiently harmonize 

 together. Keeping thus the Winteracece in the relation best 

 adapted to them, after all the other families of the Poly earpiece, 



