ANGIOSPERMS. 369 



down the length of the furrow {suture) between the two loculaments, while at the 

 same time the tissue that divides them is more or less torn and so both loculaments 

 are opened simultaneously by the longitudinal fissure (Fig, 299) ; this it was that 

 caused such anthers to be strangely called bilocular ; but they must be called 

 quadrilocular, if our nomenclature is to be scientific, in contradistinction to the really 

 bilocular anthers of the Asclepiadeae and to those of many of the Mimoseae which 

 have eight loculaments. In some cases the anther-lobes open at the apex by a pore 

 formed simply by the destruction of a small portion of tissue at the spot (Hofmeister). 

 But we still want a more detailed and comparative investigation of these processes, 

 which are of great physiological importance and at the same time very various ; here 

 we can only add the remark, that it is an important point for purposes of classification 

 whether the anther-lobes open inwards, towards the gynaeceum, or outwards, and 

 this depends on whether the pollen-sacs are on the inner or outer side of the filament. 

 More or less important deviations ^ from the course of development and from 

 the ultimate structure of the pollen as described above occur in several families 

 of Monocotyledons and Dicotyledons. In Naias and Zos/era the difference is that the 

 walls of the mother-cells are not thickened, and the pollen-grains themselves have very 

 thin walls ; the latter have a very unusual appearance in Zostera, being long narrow 

 tubes which lie parallel to each other in the anther, instead of having the ordinary 

 rounded form. There are more considerable deviations however in the formation 

 of compound pollen-grains ; either the four daughter-cells (the pollen-grains) of a 

 mother-cell remain more or less closely connected, as in the case of the tetrads (four- 

 fold grains) of some of the Orchideae (Fig. 298), of Fourcj-oya, Typha, Anona. 

 Rhododendron ; or the whole products of a primary mother-cell do not separate, but 

 form one mass of eight, twelve, sixteen, thirty-two or sixty-four pollen-grains all con- 

 nected together, as in many species of Mimoseae, Acacieae, and others^; in these cases 

 the cuticle (exine) on the free outer surface of the grains which lie at the circumference 

 of the mass is more strongly developed and covers the whole as a continuous 

 membrane, from which thin ridges only project inwards between the individual 

 cells. All gradations occur in the different divisions of the Orchideae from the 

 ordinary isolated pollen-grains of the Cypripedieae through the four-fold grains of 

 the Neottieae (Fig. 298) up to the Ophrydeae, in which all the jiollen-grains formed 

 from a primary mother-cell remain in connection, and thus a number of pollen-masses 

 {massulae) lie together in each loculament, and finally to the poUinia of the 

 Ceriorchideae, in which all the pollen-grains of a loculament are united into a 

 cellular mass. In this case, as in the Asclepiadeae with l)ilocular anthers in which 

 the pollen-grains of each loculament are held fast together by a wax-like sulisiance. 



' Hofmeister, Neue Beitr. II (Abhandl. d. K. Siichs. Ges. VII).— Reichciibach, De pollinis 

 Orchidearum genesi, Leipzig, 1852.— Rosanoff, Ueber den Pollen d. Mimoseen (Jahrli. f. wiss. Bot. 

 VI. 441).— [Corry, Struct, and Developm. of the Gynostegium &c. in Asclepias Cornuti /frans. Linn. 

 Soc. London, 1884 .] 



^ The anthers in many of the Mimoseae have eight loculaments, and a still larger number have 

 been observed \ Engler, loc. cit. en p. 354V In the Onagrarieae also, as in Gaura and others, more than 

 four loculaments occur, but the history of development requires closer investigation in these cases. 

 In the plurilocular Mimoseae the pollen-sacs are filled with groups of eight, sixteen or thirty-two 

 cells, which have originated in one primary mother- cell. 

 [2] ■ Rb 



