omitted. It is, however, worthy of
remark, that in the early stage of the ovulum, this envelope is in
general hardly visible even in those cases where, as in Hibbertia
volubilis, it attains the greatest size in the ripe seed; nor does it in
any case, with which I am acquainted, cover the foramen of the testa
until after fecundation.

The testa, or outer coat of the seed, is very generally formed by the
outer membrane of the ovulum; and in most cases where the nucleus is
inverted, which is the more usual structure, its origin may be
satisfactorily determined; either by the hilum being more or less
lateral, while the foramen is terminal; or more obviously, and with
greater certainty where the raphe is visible, this vascular cord
uniformly belonging to the outer membrane of the ovulum. The chalaza,
properly so called, though merely the termination of the raphe, affords a
less certain character, for in many plants it is hardly visible on the
inner surface of the testa, but is intimately united with the areola of
insertion of the inner membrane or of the nucleus, to one or other of
which it then seems entirely to belong. In those cases where the testa
agrees in direction with the nucleus, I am not acquainted with any
character by which it can be absolutely distinguished from the inner
membrane in the ripe seed; but as a few plants are already known, in
which the outer membrane is originally incomplete, its entire absence,
even before fecundation, is conceivable; and some possible cases of such
a structure will be mentioned hereafter.

There are several cases known, some of which I have formerly noticed,* of
the complete obliteration of the testa in the ripe seed; and on the other
hand it appears to constitute the greater part of the substance of the
bulb-like seeds of many Liliaceae, where it no doubt performs also the
function of albumen, from which, however, it is readily distinguished by
its vascularity.** But the most remarkable deviation from the usual
structure and economy of the outer membrane of the ovulum, both in its
earliest stage and in the ripe fruit, that I have yet met with, occurs in
Banksia and Dryandra. In these two genera I have ascertained that the
inner membrane of the ovulum, before fecundation, is entirely exposed,
the outer membrane being even then open its whole length; and that the
outer membranes of the two collateral ovula, which are originally
distinct, cohere in a more advanced stage by their c