SPECIES OF ZANNICHELLIA. 267 
The length of the filament offers great variations. Having studied 
these species in a dried state only, in which the stamens are rarely 
found, I am unable to appreciate with exactness the value of these 
differences. In the variety which I haye observed at Bone, the fila- 
ment was nearly three inches in length; in the form named repens, I 
have found it nearly equal to the carpel; in the major it is three 
times longer ; but M. Gay has remarked that the anther is at first 
sessile, and that the filament becomes gradually elongated, conse- 
quently I am induced to regard these differences of as little import- 
ance as those of the peduncles. 
The embryo (see figs. 10, 12, 18) is similar in all the carpels, and 
such as authors have described it, viz. the cotyledon is folded back 
against the caulicule (tigelle) after having been twice folded on it- 
self; the two latter folds, however, instead of being placed between 
the caulicule and the first portion of the cotyledon, in the same plane, 
are thrust out on the one side, so that, on examining the embryo on 
the other side, only a single fold is visible (see fig. 13), whilst the 
whole three are seen on this (see fig. 12). From the figure given by 
Reichenbach of Z. polycarpa, Nolte, we are led to imagine that the 
embryo of this species differs from that of the others, inasmuch as the 
cotyledon appears founded only on the caulicule (tigelle) ; however, 
as this is not sufficiently explained in the text, and as the figure agrees 
exactly with one of the sides of the embryo of other species, I am 
unable to fix my opinion with regard to this point. 
From all that has been said, it appears quite clear that we must not 
recognize as species sufficiently characterized all the forms which 
have been lately described, nor must we imitate M. Koch, who re- 
gards them all as mere varieties of Z. palustris, Linn. (see Syn. 
Flor. German. and Helv. t. ii, p. 679) ; but we must return to the 
species of Micheli and of Willdenow, the others appearing to be 
nothing more than sub-species or mere varieties. I shall endeavour 
to refer them to their types, for which purpose I shall be obliged to 
adopt the names of Willdenow, which, however, are not those which 
however, cause us to regard the character of the stamen as valueless ; for, 
first, it appears to be very rare, since Smith says he had never seen Zanni- 
chellia in England otherwise than with four-celled anthers; and secondly, 
the contrary never occurs, at least it is probable, since all authors inhabiting 
the interior, and having at their disposal only Z. dentata, agree in describing 
the anther as two-celled. Treviranus had searched in vain (although he had 
directed his attention to this very question) for anthers with four cells,—See 
Symb. Phyt. fase. i, Gott, 1831, 
