OF PLANTS CALLED COMPOSITE. 265
ever, connecting with it the ingenious hypothesis of M. Decandolle, namely, that petals are only modified stamina. It remains to be seen on what ground M. Cassini has adopted this theory, as proposed by M. Decandolle, for Compositse, the only family which seems to present a very important objection to it, in having its principal, and in the greater part of the order its only, vessels occupying the lines of junction of the supposed united petals.
To adapt this disposition of vessels to the theory, M. Cassini is obliged to subdivide their apparently simple trunks; a division, however, which may be regarded as entirely hypothetical. From the observations I have made on the subject, I have no doubt that these trunks are equally simple with the secondary nerves when present, or with the primary in other families. I find them to con- sist of two kinds of vessels, the spiral and ligneous. Of the spiral vessels there are generally several in the cord : in Helianthus multifiorus, however, I have not been able [83 to find more than one, either in the trunk of the nerve above the insertion of stamina, or in the branches of the latinise. It will be of some interest to verify this fact (which I by no means give with absolute confidence), both on account of the apparently formidable objection it pre- sents to the theory in question, and also that, in following- it up by an examination of the point of division, a clearer idea may be obtained of the ramification of spiral vessels than has hitherto been given.
My second objection to M. Cassini's account is, that he describes the nerves as marginal through their whole length. I have formerly, in the passage already quoted, stated them to be parallel and approximated to the margins of the latinise. Perhaps in no instance can the branches be considered as strictly marginal ; in many cases they are manifestly distinct from the margins, and in the genus Hymenopajppus are further removed from them than from the axis of the latinise. In H. scabiosceus there is also an evident inequality of the two branches in each lacinia, the stronger extending nearly to the apex, while the weaker either entirely disappears before it reaches the stronger, or
�� �
