408 MR. B. CLARKE ON THE EMBRYOS OF ENDOGENS 



cumstances, as long as the cotyledon, but no trace of it is present before germination. It 

 cannot be a rudimentary leaf, because it has no connexion with the base of the plumule, 

 1)ut only vrith the margins of the opening in the cotyledon. 



Irklece. — The embryos of several species of Lns agree in their germination mth those 

 of AracecR and Liliacece ; they are, however, remarkable for the cotyledon sending up a 

 sheatli surrounding the plumule, which sometimes becomes longer than the cotyledon 

 itself, and terminates by elongating internally between the plumule and cotyledon (fig. 26). 

 In other species of the same genus, this sheath exists only as membranous margias of the 

 elongated opening in the cotyledon, which proves beyond doubt it is of the nature of 

 stipules. 



AmaryllidecB. — The germination of Alstromeria aurantiaca agrees precisely with the 

 species of Iris in the primary radicle being exorhizal, and the secondary endorhizal, or at 

 least breaking through the cellular surface ; but the stipulary process, as I have termed 

 it, is wanting, which is a further proof of its real nature in Iris. 



Dioscoreacece. — The germination of Tainiis communis agrees most with that of the 

 Liliacece. The first leaf of the plumule alternates with the cotyledon, and not only is it 

 exorhizal, but it is some time before any distinction between the cotyledon and radicle is 

 apparent. The opening for the emission of the plumule is horizontal, and a sheatli, like 

 that which forms in Asparagus, rises round the nascent plumule, and, as growth advances, 

 becomes external to it instead of next the cotyledon, which gives it in some degree the 

 appearance of a second cotyledon ; it is, however, often bifid, and the young leaves, as well 

 as those which are mature, have lateral stipules, which alone might be sufl&cient to explain 

 its true character *. 



With reference to the value of the characters, it may be observed that they will probably 

 prove of some practical use in subdividing the Endogens; Graminacece, Zosteracem, 

 Naiaclece, and their allies forming a natural section, and containing, it may be antici- 

 pated, all the truly endorhizal and macropodal hypogynous families ; and, on the other 

 hand, Aracece, I/iliacece, Dioscoreacece, Iriclece, and their allies, in which these characters 

 are wanting, are all otherwise in near afiinity to each other. 



Such a subdivision, however, would involve the placing of the Marantacece, Musacece 

 and Amomece, and also the Orchidece and Burnianniacece on account of their large 

 radicle, in the endorhizal section, with which they have scarcely any connexion except 

 the affinity between Orchidece and Triuridece ; but yet their epigynous character should 

 not perhaps be regarded as excluding them, because of the truly epigynous character of 

 the Hydrocharidece. If, however, Von Martins is right in his comparison of Burmanni- 

 acece with Hi/drocharidece, to Avhich he refers them, another link of connexion between 

 Orchidece and the hypogynous endorliizal families may be traced ; and it may also deserve 

 notice, that the embryo of Orchis 3Iorio in germination becomes in shape much like that 

 of a Buppia, the young plumule also springing from the same part. The ovules of Bur- 

 manniacece are orthotropal, and the distichous inflorescence of Burmannia dasyanthai 



* This stipulary process occurs in the ungerminated embryo of Dioscorea cordifolia, as figured by Ad. de Jussicu, 

 and is represented as bifid. (Ann. des Sciences Nat. 2°" se'rie, xi. Bot. pi. 17. fig. 13.) 

 t Nov. Geu. et Sp. Plant. Brasil. torn. i. tab. .). fig. ' 



