It was, therefore, a real advantage to the flowers to narrow their circle of visitors more and more by varying so that only the useful visitors could gain access to their nectar, and that the rest should be excluded. Thus there arose 'bee-flowers,' 'butterfly-flowers,' 'hawk-moth flowers,' and, indeed, in many cases, a species of flower has become so highly specialized that its fertilization can only be brought about by a single species of insect. This explains the remarkable adaptations of the orchids and the enormous length of the proboscis in certain butterflies. Even our own hawk-moths _Macroglossa stellatarum_ and _Sphinx convolvuli_ show an astonishing length of proboscis, which measures 8 cm. in the latter species. In _Macrosilia cluentius_, in Brazil, the proboscis is 20 cm. in length; and in Madagascar there grows an orchid with nectaries 30 cm. in length, filled with nectar to a depth of 2 cm., but the fertilizing hawk-moth is not yet known. Thus we may say that the flowers, by varying in one direction or another, have selected a definite circle of visitors, and, conversely, that particular insect-groups have selected particular flowers for themselves, for those transformations of the flowers were always most advantageous which secured to them the exclusive visits of their best crossing agents, and these transformations were, on the one hand, such as kept off unwelcome visitors, and, on the other hand, such as attracted the most suitable ones. From the botanical point of view the assumption that flowers and flower-visiting insects have been adapted to each other by means of processes of selection has been regarded as untenable, because every variation in the flower presupposes a corresponding one in the insect. I should not have mentioned this objection had it not come from such a famous naturalist as Nägeli, and if it were not both interesting and useful in our present discussion. Nägeli maintained that selection could not, for instance, have effected a lengthening of the corolla-tube of a flower, because the proboscis of the insects must have lengthened _simultaneously_ with it. If the corolla-tube had lengthened alone, without the proboscis of the butterfly being at the same time elongated, the flower would no longer be fertilized at all, and if the lengthening of the proboscis preceded that of the corolla-tube it would have no value for the butterfly, and could not therefore have been the object of a process of selection.