176 
OKGANOGRAPHY. 
BOOK I. 
the connective divides into two unequal portions, one of which 
supports a cell and the other is cell-less ; in this case the 
connective has been called by Richard, distractile. Lacistema 
(Plate IV. fig. 7.) affords another instance of a divided con- 
nective. In many of the cases of excessive divergence of the 
cells the line of dehiscence of the anther is changed from lon- 
gitudinal to vertical (Plate III. fig. *20. 17.), and has actually 
been supposed to be really transverse; an error which in most 
cases has arisen from not understanding the real structure of 
the anther. Some anthers, however, no doubt have cells that 
burst transversely, as Lemna, Alchemilla arvensis, Securi- 
nega, &c. (See Plate III. fig. 12. 16. 30.) 
All anthers are not two-celled, their internal structure 
being subject to several modifications. It sometimes happens 
that the anther is four-celled, as in Tetratheca. In Epacris 
the two ordinary cells become confluent into one, and the 
anther is therefore one-celled. In Maranta and Canna only 
one cell is produced, the other being entirely suppressed. In 
most Aniarantacese, and some other plants, the anther seems 
to be absolutely one-celled. (Plate IV. fig. 8.) 
Of all these the four-celled anther is the type, and both the 
one-celled and two-celled are probably mere modifications of 
it, depending upon whether the septa which originally exist 
all remain complete, or are half absorbed, or wholly absorbed. 
Schleiden says he has found the anther before its bursting 
quadrilocular in more than one hundred families; amongst 
which may be named Graminaceae, Cyperaceae, Liliaceae, La- 
biatae, Borraginaceae, Scrophulariaceae, Compositae, Apiaceae, 
Ranunculaceae, with its allies, Rosaceae (Juss.), and Legumi- 
nosae, which orders alone constitute almost one half of the 
entire vegetation of the globe. It has been often asserted, 
he says, that the anther could not originally be quadrilocular, 
because it opens by two fissures only ; which is as much as to 
consider two rooms in a house as one, because they have not 
folding doors, but single doors placed close together. Pro- 
perly speaking, every anther really opens with four fissures ; 
they appear, however, only as two, because each pair lies at 
the side of the common septum. 
Other deviations from the normal form of anther occur. 
