IN JAVA. 



87 



narrow channel by which the viscid matter of the stigma 

 reaches the anther. In esainining an adranced bnd, the viscid 

 matter of the stigma is seen t^ be in large quantity and rather 

 liquid. It increases with the growth of the flower till it 

 overflows, — often before the bud opens — and, immediately on 

 its opening, inundates the poll in ia, which now increase in size, 

 and either avalanche downwards, sometimes quite obliterating 

 the rostelhmi (Fig. 2, p. 86) ; or, while retaining their position 

 in the anther, emit their tubes over the narrower portion of 

 the rostelliim into the stylary canal. Very often both anther 

 and stigma become quite filled up by the multitude of polleo- 

 tubes and by the swollen pollinia. All these plants produced 

 large and well-filled seed-capsules on every flower j but I 



no. 5.— PHAjrs blpmei, SBOwrao thk fig. 6.— t'HAJtrs blumei, siiowjnh a 



ANTHER BOTATEO EOTOfWABDa ] A, MOEK ADVANCED HTAOB THAX FIG. i) ; 



AS IN FIG. 3, THE ANTHER-GAP A, HAS OPENKD ; 



B, BWOt,!^ POLLUVIA ; C, E, AS IN 

 1*1 0. 2 ; tTF OP AKTIIEK-CAP- 



never saw an insect visit the plants during all iny observations, 

 although the plants were situated where I could inspect them 

 constantly throughout the day or night. 



Of flowers of the second form, I examined many more examples. 

 Here there is no rostellum, nevertheless the boundaries of the 

 stigma are quite distinct (Figs. 3, 4, p. 86). On examining a 

 young hid, the anther (enclosing the pollinia) is seen standing 

 vertically erect on the top of the column— i.e. of the detached 

 column, without reference to its position in the flower — forming 

 as it were a pointed extension of it, and attached to it by its 

 minute ^lament As the flower progresses in growth,the anther- 

 cap ruptures and rotates forward. When it has descended 

 through about 90% it occupies (Fig, 5) the position which, if it 

 possessed a rostellum, it would naturally retain ; but, having 



