124 Mr.. G. BENTHAM OX MYrvTACEiE. 



between the typical ones. The embryonic charactez' in the fleshy- 

 fruited Mj-rtese is therefore more artificial than was supposed, and 

 is only retained for want of a better one. Myrcia^ indeed, can 

 generally be known by the ovules, two only in each cell, whilst 

 they arc more numerous in Myrtiis and in Evgenia ; but the two 

 latter genera, when the peduncles are 1-flowered, have nothing 

 but the embryos to separate them. When several-flowered, the 

 American Eugeuias are racemose, tlie Myrcias and Myrtuscs cy- 

 mose, as well pointed out by Grisebach. The Asiatic and African 

 many-flowered Eugenias are, it is true, cymose ; but there are there 

 no Myrcias or many-flowered Myrtuscs to confound with them. 

 The Australian Eugenias are chiefly of the Asiatic type ; but there 

 are some species oi Myrtus^'iih the inflorescence of the American 

 Eugenias, obliging us thus to depend solely upon the embryonic 

 character irrespectively of geographical origin. 



In Lecythideae the diversity of embryo, in the ten genera in 

 which it is known, is as great as in either of the other tribes. In 

 Barrinrjtonia and Careya it is a hard, thick, undivided mass, with 

 a line down the centre, which had been supposed to be the indi- 

 cation of a separation between two cotyledons, until it w^as shoAvn 

 by Thomson (Journ. Liun. Soc. ii. 47) to be of a pithy nature, 

 and that the real cotyledons are abortive. In Gustavia and 

 J^poleona the embiyo, as in JBarringtonia^ is thick and hard, or 

 fleshy, filling tbe seed, but, as in Eugenia,- it consists of two 

 distinct cotyledons with a small radicle. In Lecythis and Ber- 

 tliolletia, again, we have the same thick mass, T)ut without any 

 indication, as far as hitherto observed, whether it is all radicle 

 without cotyledons, as in Barringtonia^ or almost all cotyledons 

 with a very small radicle, as in Gustavia, In Planclionia, other- 

 wise closely allied to Gustavia, and in Lecythopsis^ otherwise 

 closely allied to LecytMs, as well as in Couratari and Courotipita^ 

 the cotyledons are much folded, and surrounded by a very long, 

 folded or spiral radicle. In Betersia^ Grias, and AsterantJios the 

 seeds are unknov\Ti. 



AVhilst, therefore, the embryo in Mjrtacese still retains a very 

 high position in the scale of generic characters, we find that when 

 we rely vipon it as absolute, as we are compelled to do sometimes 

 for want of a better one, our genera may become artificial, and 

 that in some cases, as in Melaleuca, we are obliged to place it 

 below staminal and some other characters. 



