8AXIFRAGAGEM. 389 



and arched at both ends, by single flaps 1 or valves, not double as in 

 Loropetalum. Between the stamens are five glandular bifid or bilobate 

 bodies, forming a sort of disk ; they are probably staminodes. The 

 ovary and ovules behave as in Hamamelis. The fruit is a bicuspidate 

 capsule, with two bifid valves, and seeds of Hamamelis. Cori/lopsis 

 inhabits temperate Central and Eastern Asia ; it comprises three or 

 four frutescent species, 2 sometimes cultivated in this country. The 

 leaves are caducous, 3 with large caducous stipules. The flowers 

 come out before the leaves, at the beginning of the season ; forming 

 pendant racemes or catkins, axillary to bracts or scales which are 

 only the stipules of aborted leaves. 



Dicort/p/ie 4 has usually tetramerous flowers, probably all hermaphro- 

 dite. The concave receptacle lodges the inferior ovary, as in the last 

 genus, but the form of the perianth is quite peculiar. The calyx 

 is a cylindrical coriaceous tube, with four valvate teeth, and comes 

 off at the base in a single piece. The petals are four thick fleshy 

 tongues. There are eight stamens, but the alternipetalous set are 

 sterile. The four others 5 have flattened elongated basifixed anthers, 

 with two introrse cells; each of these opens by a half-valve, 6 correspond- 

 ing with half its wall, or by the opening of the whole of the outer wall 

 into a complete valve. There are two cells to the ovary; the ovules, 

 originally two in each cell, behave exactly as in Hamamelis. The 

 fruit is a capsule. This genus comprises five or six shrubs from 

 Madagascar, 7 with alternate or opposite entire persistent coriaceous 

 leaves, possessing unsymmetrical stipules, often large and caducous. 

 The flowers form terminal racemes, sometimes short, with the pedicels 

 so short as to simulate cajDitula. 



1 Their dehiscence shows a transition from the Noronh., mss. (ex Tul.). — Glycoxylum Chapel., 

 longitudinal cleft to the valves that are so marked mss. (ex Tul.). 



in other genera, which greatly lessens the value 5 The filaments are united in D. stipulacea, but 



of the character. we do not know whether they only stick together 



2 Griff., PI. Cantor., 22. — Hook. f. & or are really monadelphous. The sterile stamens 

 Thoms., in Journ. Linn. Soc, ii. 85. — Hance, may stick to the contracted base of the petals, 

 in Ann. So. Nat., ser. 4, xv. 224. — Bot. Mag., t. without, however, any real fusion. 



5158. — Walp., Rep., ii. 434; Ann., vii. 936. 6 In D. stipulacea the anther first opens by a 



3 Sprinkled with stellate hairs, like the young lateral cleft on each side ; then the two internal 

 branches. half-cells bend inwards towards one another, while 



4 Dup.-Th., Gen. Nov. Madag., 12 ; Hist, des the dorsal ones remain in situ. 



Yeg. des lies Afr. Austr., 31, t. 7. — DC, Prodr., 7 Jaume S.-Hil., Exp. Fain. Nat., ii. 368. — 



iv. 269. — H. Bn., in Pager Fam, Nat., 344. — RcEM. & Sen., Sgsl., iii. 845. — Tul., in Ann. So. 



Endl., Gen., n. 4588. — B. H., Gen., 667, n. 6.— Nat., ser. 4, viii. 142.— Walp., Ann., vii. 836. 

 JDicorgpha Spreng., Syst., i. 546. — Diania 



