299, METAMORPHY. 
tending to show that the anther is in such cases really 
a modification of the blade of the leaf; but as, on the 
other hand, we often find petal-lhke filaments bearing 
pollen-sacs on their sides, it is clear that we must not 
attribute the formation of pollen to the blade of the 
leaf only, but we must admit that it may be formed in 
the filament as well.’ 
1 Although it is generally admitted that the filament of the stamen 
corresponds to the stalk of the leaf, and the anther to the leaf-blade, yet 
there are some points on which uncertainty still rests. One of these is 
as to the sutures of the anther. Do these chinks through which the 
pollen escapes correspond (as would at first sight seem probable) to the 
margins of the antheral leaf, or do they answer to the lines that separate 
the two pollen-cavities on each half of the anther one from the other ? 
Professor Oliver, ‘ Trans. Linn. Soc.,’ vol. xxiii, 1862, p. 423, in alluding to 
the views held by others on this subject, concludes, from an examination 
of some geranium flowers in which the stamens were more or less petaloid, 
that Bischoff’s notion as to the sutures of the anther is correct, viz., 
that they are the equivalents of the septa of untransformed tissue 
between the pollen-sacs. Some double fuchsias (‘ Gard. Chron.,’ 1863, 
p. 989) add confirmation to this opinion. In these flowers the petals 
were present as usual, but the stamens were more or less petaloid, the fila- 
ments were unchanged, but the anthers existed in the form of a petal-like 
cup from the centre of which projected two imperfect pollen-lobes (the 
other two lobes being petaloid). Now, in this case, the margins of the 
anther were coherent to form the cup, and the pollen was emitted along a 
line separating the polliniferous from the petaloid portion of the anther. 
: This view is also borne out by the double-flowered 
Arbutus Unedo, and also by what occurs in some 
double violets, wherein the anther exists in the 
guise of a broad lancet-shaped expansion, from 
the surface of which project four plates (fig. 157), 
representing apparently the walls of the pollen- 
sacs, but destitute of pollen; the chink left be- 
tween these plates corresponds thus to the suture 
of the normal anther. 
The inner or upper portion of the anther-leaf 
is that which is most intimately concerned in the 
formation of pollen; it comparatively rarely 
(query ever) happens that the back or lower 
surface of the antheral leaf is specially devoted 
to the formation of pollen. On the other hand, 
in cases like those of the common honuseleek, 
where we meet with petaloid organs combining 
the attributes of anthers and of carpels, we find 
the inner layers devoted to the production of pollen, 
the outer to the formation of ovules. 
That the pollen-lobes are not to be taken as halves of a staminal leaf, 
but rather as specialised portions of it, not necessarily occupying half 
its surface, is shown also in the case of double-flowered Malvaceae, in 
which the stamens are frequently partly petal-like, partly divided into 

Fic. 157. Peta- 
loid stamen of 
Viola, with four 
projecting plates. 
