surface of the roll, and by the other to the flat surface of the leaf. They are so many bands, so many little cords which hold out against the spring of the leaf. There are sometimes more than from ten to twelve of these bands arranged nearly in the self-same straight line. Each band is a packet of threads of white silk, pressed one against the other, and yet we must remember all are separate."[75] [75] Mémoires pour servir à l'Histoire des Insectes, tome ii., page 210 (5e Mémoire). [Illustration: Fig. 282.--Oak leaf rolled perpendicularly.] [Illustration: Fig. 283.--Oak leaf rolled sideways.] Réaumur made the oak-leaf rollers work in his house. He has admirably described all their little manoeuvres, but we lack the space to convey to the reader the result of his minute observations. In fact, the leaf-rollers construct for themselves a sort of cylindrical cell, which receives light only through the two extremities. The convenience of this green fresh habitation is, that its walls furnish food to the animal which inhabits it. The caterpillar, thus sheltered, sets to work to gnaw away at the end of the leaf which it rolled first; it then eats all the rolls it has made, up to the very last. [Illustration: Fig. 284.--Leaf of sorrel, a portion of which is cut and rolled perpendicularly to the leaf.] Réaumur found also rolls which had been formed of two or three leaves rolled lengthwise, and he saw that the leaves which had occupied the centre had been almost entirely eaten. He saw also caterpillars which continued to eat while they were making their habitation. Let us add that one of the ends of the roll is the opening through which the caterpillar casts its excrement; that the caterpillar can prepare itself a fresh roll if it is turned out of the first; and, lastly, that it is in a rolled leaf that the caterpillar undergoes its metamorphoses into a chrysalis and into a moth. Réaumur studied other leaf-rollers; for instance, those which roll the leaves of nettles and of sorrel. The last one works in a manner which deserves to be mentioned. Its roll is of no particular shape, but it is its position which is remarkable. It is set upon the leaf like a ninepin (Fig. 284). The caterpillar has not only to twist it up into a roll, but also to place it perpendicularly on the leaf. [Illustration: Figs. 285 and 286.--Willow leaves rolled by a caterpillar, and section of a bundle of leaves drawn together by a caterpillar.]