AND THEIR MORPHOLOGY. 131 
It is visited, exclusively during the day, as far as I can see, bya 
splendid bee, probably a Euglossa, but with the tongue nearly twice 
as long as the body. The tongue passes out behind the abdomen, 
and is there curved upwards. As these also only come for biting 
and gnawing the anterior side of the labellum, the protruding 
tongue touches or approaches the gland at every retrograde 
movement of the insect. By this it. can hardly fail to be loaded 
Sooner or later with the pollen-masses, which are then easily 
inserted into the stigmatic cleft. I have, however, not as yet 
observed this fact. 
While in Catasetum one flower is always impregnated by pollen 
of another, the possibility of self-impregnation exists in the other 
three examples, and I have no doubt that it often happens. In 
Epidendree I have also noticed it many times; and I believe it is 
owing, in the latter cases, to the abundance of stigmatic viscosity 
on the face of the stigma, which is situated, in nearly all plants of 
this suborder, immediately below the pollen-bed. We have here in 
Trinidad three plants belonging to Epidendree—a Schomburgkia, 
a Cattleya, and an Epidendrum—which rarely open their flowers, 
and invariably are impregnated when they do not open them. In 
these cases it is easily seen that the pollen-masses have been acted 
upon by the stigmatic fluid, and that the pollen-tubes descend 
from the masses still Zu situ down into the ovarian canal. This 
has also been shown to be the case in a certain class of dimorphic 
flowers, as in Viola and Oxalis, where the pollen emits tubes 
from the anthers, which tubes enter the stigma and descend to 
the ovules (see H. v. Mohl, Bot. Zeitung, 1868, Nos. 42 & 43). 
But, surrounded as we are by innumerable facts demonstrating 
that self-impregnation is, contrary to what was formerly supposed, 
not the rule, and necessary self-impregnation an extremely rare 
case, I must entirely demur to the conelusion that these few facts 
are destructive to the Darwinian theory, or, as Mohl has it, are 
of equal value to prove a contrary theory. Probabilities deduced 
from the number of observed facts must always enter for a large 
part into our theories, in sciences of a complex nature. It is true 
that a complete theory admits of no exceptions ; but nobody will, I 
believe, maintain that the above theory has arrived at that state. 
A few residual facts will not disturb our admiration for it, and the 
harmony into which it has brought so many branches of natural 
history hitherto unconnected. As far as intercrossing, and the 
gradual variation and transformation arising therefrom especially 
are concerned, there is no necessity to represent to ourselves 
