MAQN0LIACE2E. 165 



to Maf/noliacem by Bentham & Hooker. Still later, Miehs' proposed 

 to put the CanellecB near the Winteracea, which included Illicium 

 and Brimi/s. The old genus Canella was at the same time split up 

 by him to permit the establishment of his genus Cinnamodendron. 

 SiEBOLD^ had in 1S35 described Trochodendron, which he put near 

 MagnoUacece. Bentham & Hooker' had made it an abnormal Araliad. 

 But Hooker & Thomson' were decided by tlie arguments of 

 Eichler/ and the comparison they were able to make with Euptelea,^ 

 another Japanese genus, at one time referred to Ulmacece, to restore 

 the two last-named genera to Magnoliace<s. Thus was raised to 

 nine the number of genera, which according to us should form 

 part of the order. We have added two others, Zygogynim^ a 

 Drimyd with a syncarpous ovary, and Cinnamosma^ a gamopetalous 

 Canellad. 



Of all the characters presented by plants of this order, there are 

 only three absolutely constant, and it must be owned that even 

 these possess but little value of themselves : they are, woody stem, 

 alternate leaves, and albuminous seeds. We can conceive that some 

 time or other a Magnoliad might be found wanting any of these 

 characters, and yet such that we could not on that account exclude 

 it from this order. But beside these absolute characters, we have a 

 very large number of others so general tliat their extremely rare 

 absence (often observed in a single genus only) is sufficient to de- 

 termine an important tribe or genus. Hence it is on these almost 

 constant characters that we must lay stress. Eight may be 

 enumerated — 



1. The form of the floral receptacle. — This, so important on account 

 of the mode of insertion which directly results from it, is more or less 

 concave in the two genera Euptdea and Trochodendron only, and 

 especially in the latter ; it suffices to characterize the series of Eiqjtelecs 

 or Trochodendrea. 



2. Again, these two genera alone lack a true perianth ; the absence 

 of calyx and corolla is an equally good characteristic of this series. 



' Contrih.,\.\\2. Flora (1864), 449 j (1865), 12 j Seem., Joum. 



- Fl. Jap. Fam., 133. of Bat., iii. (1865), 150. 



^ Gen., 17. * Established by SiEBOlD&ZrccAElNl in 1835. 



* Joum. Linn. Soc, vii. (1863), 240. ' In 1867 ; see p. 156, note 3. 



* Mart. Flor. Bras., MagnoUac, 131 ; * In the same year ; see p. 163, note 1. 



