1G4 NATURAL HISTORY OF PLANTS. 



with the mioropyles looking downwards and inwards. The fruit is 

 a many-seeded berry,' but the most striking character of this genus is 

 its gamopetaloas corolhi, whose tube elongates as it grows older, and 

 whose limb, first spreading and afterwards reflexed, is divided either 

 into five quincuncial lobes, whose position with regard to the sepals 

 is the same as in Canella, or in six lobes, of which three are 

 external and three internal. C. fragrans, the only species known, is a 

 small tree with alternate exstipulate aromatic leaves, which as j^et has 

 only been observed in the north of Madagascar. It may be defined 

 as a Canella, with solitary axillary sessile flowers and a gamopetalous 

 corolla. 



Havinor ascertained and discussed the characters of the eleven 

 genera that we retain in this order, let us now see how each, in suc- 

 cession has come to be placed in it. B. de Jussteu' had. ranked in 

 his TlUciP the Magnoliacca properly so called — i.e., the genera Lirio- 

 dendrum and Magnolia. lUicimn alone was placed among his Anona. 

 Adanson',^ far more logical, included in one and the same order the 

 Anona, the genera Illicium (under the nsiViiQ oi Skm mi), Magnolia, 

 Champaca {Michelia), and Tulipifera. As we also find Dillenia and 

 Menispermian in this family, it is evident that this genius left nothing 

 of the true aflfinities of MagnoliacecB to be discovered by modern 

 botanists. A. L. de Jussieu* had simply to divide the Anonacece of 

 Ad.\nson into two nearly equal parts : he separated Anona and several 

 nearly allied genera to constitute his order Anona, and left as true 

 MagnoliacecB the genera Drimys, Illicium, Michelia, Magnolia, Talauma, 

 and Liriodendrum. Unfortunately, to these he added Ei/rgandra 

 {Tetracera) and Magna, and also Billenia, Curatella, Ochna, and 

 Quassia, as "genera a^nia." Still, thenceforward four of the 

 genera which we retain as distinct in MagnoliacecB were united into 

 one group. Canella was placed among the MelicB. The genus 

 Schizandra, taken by Blume* as the type of a separate order, Schizan- 

 dracecB, retained as distinct by authors until 18()2,^ was then referred 



' The seeds are surromuled with the same ' Fam. des Plant., ii. 364. 



son of puip US that which is so abundant in Cin- * Gen., 280, ordo xv. 



namodendron . Tliey are probably of similar ^ Bijdraj. (1825), 21. 



Btructure to those of Canella, but have not yet * Exdl., Gen., 835. — Meisnee, Gen., 5. — 



been studied when quite ripe. Lixdl., Veget. Kingd., 305. 



' Ex A. L. Juss., Oen., Ixviii. 



