Ewart. — On the Staminal Hairs of Thesium . 275 
a slight prominence above the perianth-surface, and consisting 
of from twelve to sixteen cells side by side. The secretion is 
very much the same as before, but the globules seem more 
finely divided. The filamentous portions spring from the 
upper end of the basal cushions and are much thinner than in 
the preceding cases ; they are quite filled with small globules 
of the secretion and extend almost to the top of the anther ; 
the free ends are constricted three or four times (Fig. 5). The 
terminal caps break off, and in one case I noticed that the 
cap had not entirely separated from the hair, but remained 
attached at one side (Fig. 5, iii). 
T. paniculatum . — The staminal hairs resemble those of T. 
dehile but are more numerous, and still longer and thinner, 
passing up behind the anther and over the top. They are 
arranged in five groups, one behind and just above each 
stamen. The secretion is abundant, the globules being very 
finely divided. In a longitudinal section of the perianth the 
basal cushions appear somewhat square and two or three 
deep ; in a transverse section they are flattened from side to 
side, and there are from twelve to fourteen in a row. The 
free ends show three constricted zones and a rounded terminal 
cap, as before, but they are not so conspicuous, owing to the 
narrowness of the hairs. The hairs are situated directly above 
the xylem of the vascular bundle which runs up the centre of 
each perianth-segment. 
T. alpinum . — The perianth is much elongated and only 
4- fid; consequently there are only four groups of staminal 
hairs, which generally resemble those of T. debile and panicu- 
latum , but are more elongated and slender; the filamentous 
portions of the hairs pass over the tops of the anthers and 
bend over to touch the stigma (Fig. 9). Their free ends 
are constricted in the usual manner. Owing to their upward 
direction the filamentous portions are necessarily cut off in a 
transverse section of the flower, such as that shown in Fig. 10. 
I have therefore represented them by dotted lines, in order 
to indicate their length and breadth, as if they had been bent 
forwards into a horizontal position. 
