It feeds upon the elm, using the leaves both as food and clothing. To do this it only eats the parenchyma of the leaf, preserving the upper and under epidermal membranes, between which it then insinuates itself as it progressively devours the parenchyma. It, however, carefully avoids separating these membranes where they unite at the extreme edge of the leaf, which is designed to form 'one of the seams of its coat.' The cavity when thus excavated between the two epidermal membranes is then lined with silk, made cylindrical in shape, cut off at the two ends and all along the side remote from the 'seam,' and then the two epidermal membranes sewn together along the side where they have had to be cut in order to separate them from the tree. The larva now has a coat exactly fitting its body, and open at each end. By the one opening it feeds, and by the other discharges its excrement, 'having on one side a nicely jointed seam--that which is commonly applied to its back--composed of the natural marginal junction of the membranes of the leaf.' Réaumur cut off the edge of a newly finished coat, so as to expose the body of the larva at that point. The animal did not set about making a new coat _ab initio_, as we might expect that it would on the popular supposition that a train of instinctive actions is always as mechanical as the running down of a set of cog-wheels, and that wherever a novel element is introduced the machinery must be thrown out of gear, so that it cannot meet a new emergency of however simple a character, and must therefore re-start the whole process over again from the beginning. In this case the larva sewed up the rent; and not only so, but 'the scissors having cut off one of the projections intended to enter into the construction of the triangular end of the case, it entirely changed the original plan, and made that end the head which had been first designed for the tail.' Another remarkable case of the variation of instinct in the Lepidoptera is stated by Bonnet. There are usually, he says, two generations of the Angoumois moth: the first appear in early summer, and lay their eggs upon the ears of wheat in the fields; the second appear later in the summer, or in the autumn, and these lay their eggs upon wheat in the granaries; from these eggs there comes the first generation of next year's moths. This is a highly remarkable case--supposing the facts to be as Bonnet states; for it seems that the early summer moths, although born in the granaries, immediately fly to the unreaped fields to lay their eggs in the standing corn, while the autumn moths never attempt to leave the granaries, but lay their eggs upon the stored wheat.[103] Westwood says that--