cases in which the upper surface of the rostellum is viscid, as in Malaxis and some Epidendrums, and simply adheres to the pollen-masses; for these cases present no difficulty. But I refer to the so-called congenital attachment of the pollinia by their caudicles to the rostellum or viscid disc. It is not, however; strictly correct to speak of congenital attachment, for the pollinia are invariably free at an early period, and become attached either earlier or later in different Orchids. No actual gradation is at present known in the process of attachment; but it can be shown to depend on very simple conditions and changes. In the Epidendreæ the pollinia consist of a ball of waxy pollen, with a long caudicle (formed of elastic threads with adherent pollen-grains), which never becomes spontaneously attached to the rostellum. In some of the Vandeæ, as in Cymbidium giganteum, on the other hand, the caudicles are congenitally (in the above sense) attached to the pollen-masses, but their structure is the same as in the Epidendreæ, with the sole difference, that the extremities of the elastic threads adhere to, instead of merely lying on, the upper lip of the rostellum.
In a form allied to Cymbidium, namely, Oncidium unguiculatum, I studied the development of the caudicles. At an early period the pollen-masses are enclosed in membranous cases, which soon rupture at one point. At this early period, a layer of rather large cells, including remarkably opaque matter, may be detected within the cleft of each pollen-mass. This matter can be traced as it gradually changes into a translucent substance which forms the threads of the caudicles. As the change progresses, the cells themselves disappear. Finally the threads at one end adhere to the waxy pollen-masses, and at the other end
