THK INITIATION OF STAMINAL ZYGOMORPHY. 699 



At the top of the filament, and close to the loculus, the cells are small and somewhat 

 compacted. A central conductive strand is observable, and can be traced up into the 

 connective. Between the upward continuation of the conductive strand and the inner 

 wall of the loculus lies a plate of dense-looking cells. These are continued into the 

 beak which has been formed at the summit of the anther. The tetrad stage has been 

 reached in the spore-mother-cells, and the outer wall of the anther is three or four 

 layers in thickness. 



The stamen-filament represented in fig. 34 is, in its ground-tissue, about thirty-three 

 cells in length. Of these, about twenty-six are of relatively large size. The remainder 

 are very minute, and form part of the almost uniformly small-celled upper portion of 

 the filament. It is to be noticed that further diff'erentiation has taken place in the 

 connective and stamen beak. In the loculus the tapetal cells still p.ersist, and the 

 pollen-grains — many of which are in the binucleated condition — are obtaining their 

 thick walls. The outer wall of the anther consists of a narrow-celled epidermis, 

 beneath which is a layer of relatively large cells. Between these and the crushed layer 

 which was the outer layer to the tapetum is a small-celled layer. 



A filament, fifty-seven or fifty-eight cells in length, is depicted in fig. 20. The 

 cell-crowding which was so marked at the top of the filament in the stage shown in 

 fig. 34 no longer exists. Sliding growth has taken place in the vascular strand which 

 can be traced directly into the connective. In the loculus only a few tapetal cells 

 remain, and no alteration is observable in the condition of the anther- wall. 



Figs. 20 and 35 are drawn to the same scale. The filament represented in fig. 35 

 is more than twice as long as that' shown in fig. 20, but on counting the cells in the 

 longitudinal rows it will be found that they number about fifty-eight, which was the 

 number previously recorded for fig. 20. No tapetal cells are figured in the loculus, 

 and an examination of the anther- wall will show that the fibrous layer has been 

 established (fig. 36). The length of the filament in this figure falls short of 4 inches, 

 and when it is remembered that the drawing is twenty-four times the size of the actual 

 object, it will be evident that the filament measures about ^ inch. That is to say, that 

 the main details of the anther were apparently completed while the filament was 

 very short. 



Fig. 37 represents an almost fully grown stamen in section. 



The anther is, to all appearances, identical in its details with that shown in fig. 35. 

 The vascular strand of the filament is well diff"erentiated, and it will be found, on 

 counting the ground-cells in the rows of which the filament is composed, that they 

 number between fifty -eight and sixty cells. 



The filament itself is about ^ inch long, but an increase in length takes place at 

 dehiscence. 



The results obtained from many flowers were in the main identical with those here 

 described, and as it was found that all members of the androecium were — for all 

 practical purposes — of the same size and form, and in the same condition at any point 



TRANS. EOY. SOC. EDIN., VOL. XLIX. PART III. (NO. 12) 94 



