Now when there are three lower doors in this way the one which is in use is the largest, and the door lying nearest to this one the next in size, while the hindmost is the smallest of all. But though those abandoned doors are now too small to fit the existing tube, they did so, no doubt, in their day, for they are exact copies in miniature of the ordinary horse-shoe shaped lower doors. The lower door actually in use may sometimes be found to have two separable cases of thick silk enclosing the central mass of earth, and this also, very probably, represents enlargement. In the nests of _N. meridionalis_ I have never found any of these abandoned doors behind the one in use, nor should I expect to find any, for if they were present they would permanently obstruct the entrance from the main tube to the branch. It is clear that it is better economy on the part of the spider to enlarge its nest rather than build a new one each time. If we compare the infant spider and its nest (fig. B, Plate IX., p. 98) with the full-grown creature and its nest (fig. A, Plate IX.), it becomes evident that the growing spider must either construct many nests of intermediate size, or frequently enlarge the original domicile. And we do in fact find nests of all sizes between the two extremes. I cannot help thinking that these very small nests, built as they are by minute spiders probably not very long hatched from the egg, must rank among the most marvellous structures of the kind with which we are acquainted. That so young and weak a creature should be able to excavate a tube in the earth many times its own length, and know how to make a perfect miniature of the nest of its parents, seems to be a fact which has scarcely a parallel in nature. When we remember how difficult a thing it is for even a trained draughtsman to reduce by eye a complicated drawing or model to a greatly diminished scale, we must own that the performance of this feat by a baby spider is so surprising as almost to exceed belief. And yet even the most complicated form of nest--namely, that of the branched double-door type--is perfectly reproduced in miniature by these tiny architects, with the upper door, lower door, main tube, and branch (fig. B, Plate IX., p. 98). In order to test whether the doors are enlarged or not I measured the surface doors of seven double-door nests and one minute cork door on April 30th, making a careful plan of the terrace wall in which they lay, in order to make sure of finding them again on my return to Mentone in October. The following table will show that all were enlarged, the average rate of increase being 1-7/10 lines in the five and a half months which had elapsed:--