384 DE. A. E. EENDLE—SYSTEMATIC 



will then correspond, as Magnus suggested, with the cup-like envelope in ZannicJiellia^ 

 which on Campbell's interpretation becomes a spathe surrounding an inflorescence, as in 

 the Aroids. Najas graminea is exceptional in the absence of the spathe surrounding the 

 male flower. 



N. marina is dioecious ; the other species in which both male and female flowers are 

 known are monoecious. The few^ specimens of N. lacerata which I have seen bear only 

 male flowers ; the female and fruit are unknown. 



AVhen mature the flow-ers generally stand singly in the fertile sheath-axil, more rarely 

 several together, a fertile shoot Avith suppressed internodes standing in the sheath-axil 

 (PI. XL. fig. 71). They are generally 2 to 3 mm. long. The male, as already indicated, 

 consists of a sessile or subsessile anther closely surrounded by a thin membranous sac-like 

 perianth, which ends above the anther in two thickened lips. The anther is more or 

 less ellipsoidal or oblong in shape, has a delicate wall of two cell-layers, and is generally 

 4-, more rarely 1-locular. The cells arc crowded with pollen-grains, oval or roundish in 

 shape, both sometimes occurring in one anther, as in A. graminea (PL XLII. fig. 197) ; the 

 grain has a single delicate uncuticularized wall, more or less filled with dark cell-contents, 

 consisting largely of starch-granules. Campbell states that in N. flexilis the microspore 

 divides into a larger vegetative and a smaller generative cell, w^hich are separated by 

 a membrane. The generative cell subsequently divides again. 



In iV. graminea, where the spathe is absent, the perianth forms tw^o large ear-shaped 

 lips above the anther (PL XLII. fig. 197). 



The thin membranous spathe conforms below^ to the outline of the flower, but is 

 prolonged above it into a cylindrical neck, which ends in a few of the characteristic 

 spine-cells. A very short peduncle may be developed below the spathe; in N. podostemon 

 it is almost as long as the flower. Before dehiscence of the anther the stalk elongates, 

 pushing the anther, still closely enveloped by the perianth, through the spathe, which 

 becomes split, sometimes laterally, sometimes from the apex dow^nward. The lips of 

 the perianth separate and the anther dehisces apically. (See Pis. XXXIX.-XLI. 

 figs. 41, 51, m, 94, 112, 120, 112.) 



The female flow^ers are naked, consisting of a more or less ellipsoidal ovary produced 

 into a narrow style, w^hich divides into 2, more rarely 3, equal or unequal, linear-tapering 

 branches provided with stigmatic papillse. The single, erect, anatropous ovule almost 

 fills the chamber, from the roof of w^hicli, just below the style, spring numerous 

 elongated, papilla-like cells, which doubtless serve to conduct or nourish the pollen-tube ; 

 similar cells are also found at the base of the ovary just in front of the micropylc. 



In a few tropical Old- World species the female flower is enveloped in a spathe closely 

 resembling that of the male (PL XL. fig. 67). 



As the flowers are always completely submerged, pollination must be effected by 

 passive falling, by currents of water, or the transport of the pollen by aquatic animals. 

 There is no evidence of the last-named. Magnus has observed the grains in N. marina 

 to germinate before leaving the anther, the wall growing out into a long pollen-tube. 



Jonsson (in Lunds XJnivers. Ars-skrift, xx.) says that as the male flowers in 

 monoecious forms stand higher on the shoot than the female wliich are mature at the 



