the midribs of the two lower sepals to the two lower stigmas, which are sometimes quite distinct, and then to look at the third group of vessels running from the base of the mid-rib of the upper sepal to the rostellum, which occupies the exact position of a third stigma, and doubt its homological nature. There is every reason to believe that the whole of this upper stigma, and not merely a part, has been converted into the rostellum; for there are plenty of cases of two stigmas, but not one of three stigmatic surfaces being present in those Orchids which have a rostellum. On the other hand, in Cypripedium and Apostasia (the latter ranked by Brown in the Orchidean order), which are destitute of a rostellum, the stigmatic surface is trifid.
As we know only those plants which are now living, it is impossible to follow all the gradations by which the upper stigma has been converted into the rostellum; but let us see what are the indications of such a change having been effected. With respect to function the change has not been so great as it at first appears. The function of the rostellum is to secrete viscid matter, and it has lost the capacity of being penetrated by the pollen-tubes. The stigmas of Orchids, as well as of most other plants, secrete viscid matter, the use of which is to retain the pollen when brought to them by any means, and to excite the growth of the pollen-tubes. Now if we look to one of the simplest rostellums,—for instance, to that of Cattleya or Epidendrum,—we find a thick layer of viscid matter, not distinctly separated from the viscid surface of the two confluent stigmas: its use is simply to affix the pollen-masses to a retreating insect, which are thus dragged out of the anther and transported to another flower, where they are retained by the almost equally viscid stigmatic surface. So that the office of the rostellum is still to
