Of the authors referred to, it may be remarked, that those who have most particularly attended to the ovulum externally, have not always examined it at a sufficiently early period, and have confined themselves to its surface:  that those who have most minutely examined its internal structure, have trusted too much to sections merely, and have neglected its appearance externally:  and that those who have not at all examined it in the early stage, have given the most correct account of its surface.  This account was founded on a very limited observation of ripe seeds, generalized and extended to the unimpregnated ovulum, in connexion with an hypothesis then very commonly received:  but this hypothesis being soon after abandoned, their statement respecting the ovulum was rejected along with it.

In the ovulum of Kingia, the inner membrane, with relation to the external umbilicus, is inverted; and this, as I have already observed, though in direct opposition to M. Turpin’s account, is the usual structure of the organ.  There are, however, several families in each of the two primary divisions of phaenogamous plants, in which the inner membrane, and consequently the nucleus, agrees in direction with the testa.  In such cases the external umbilicus alone affords a certain indication of the position of the future embryo.

It is an obvious consequence of what has been already stated, that the radicle of the embryo can never point directly to the external umbilicus or hilum, though this is said to be generally the case by the most celebrated carpologists.

Another observation may be made, less obviously a consequence of the structure described, but equally at variance with many of the published accounts and figures of seeds, namely, that the radicle is never absolutely enclosed in the albumen; but, in the recent state, is either immediately in contact with the inner membrane of the seed, or this contact is established by means of a process generally very short, but sometimes of great length, and which indeed in all cases may be regarded as an elongation of its own substance.  From this rule I have found one apparent deviation, but in a case altogether so peculiar, that it can hardly be considered as setting it aside.

It is necessary to observe, that I am acquainted with exceptions to the structure of the ovulum as I have here described it, In Compositae its coats seem to be imperforated, and hardly separable, either from each other or from the nucleus, in this family, therefore, the direction of the embryo can only be judged of from the vessels of the testa.* And in Lemna I have found an apparent inversion of the embryo with relation to the apex of the nucleus.  In this genus, however, such other peculiarities of structure and economy exist, that, paradoxical as the assertion may seem, I consider the exception rather as confirming than lessening the importance of the character.

(Footnote.  Linnean Society Transactions 12 page 136.)