Fourth experiment.—It is almost impossible to be certain whether S. flavipennis, which makes several calls at the bottom of the same passage, and heaps several grasshoppers in each, commits the same irrational mistakes when accidentally disturbed. A cell may be closed, although empty or imperfectly stored, and yet the Sphex will return to the same burrow to make others. Yet I have reason to believe that this Sphex is subject to the same aberrations as her two relations. The facts on which I base my belief are these. When the work is completed, there are generally four grasshoppers in each cell, but it is not uncommon to find three or only two. Four appears to me the usual number—first, because it is the most frequent, and secondly, when I have brought up young larvæ dug up when eating their first grasshopper, I found that all, even those only provided with two or three, easily finished those offered, up to four, but after that they hardly touched the fifth ration. If four grasshoppers are required by the larva to develop fully, why is it sometimes only provided with three or even only two? Why this immense difference in the amount of food? It cannot be from any difference in the joints served up, since all are unmistakably of the same size, but must come from losing prey on the road. In fact, one finds at the foot of the slopes whose upper parts are occupied by Sphegidæ, grasshoppers killed, and then lost down the incline, when, for some reason or other, the Sphex has momentarily left them. These grasshoppers become the prey of ants and flies, and the Sphex who finds them takes good care not to pick them up, as they would take enemies into the burrow. These facts seem to demonstrate that if S. flavipennis can compute exactly how many victims to catch, she cannot attain to counting how many reach their destination, as if the creature had no other guide as to number than an irresistible impulse leading her to seek game a fixed number of times. When this number of journeys has been made,—when the Sphex has done all that is possible to store the captured prey,—her work is done, and the cell is closed, whether completely provisioned or not. Nature has endowed her with only those faculties called for under ordinary circumstances by the interests of the larva, and these blind faculties, unmodified by experience, being sufficient for the preservation of the race, the animal cannot go farther. I end then as I began: instinct knows everything in the unchanging paths laid out for it; beyond them it is entirely ignorant. The sublime inspirations of science, the astonishing inconsistencies of stupidity, are both its portion, according as the creature acts under normal conditions or under accidental ones. XIII AN ASCENT OF MONT VENTOUX