In a tumbler small enough to allow of easy and precise observation, I install three or four caddis worms, extracted this moment from their sheaths with every possible precaution. After a number of attempts which have at last shown me the right road, I place at their disposal two kinds of materials, possessing opposite qualities; the supple and the firm, the soft and the hard. On the one hand, we have a live aquatic plant, such as watercress, for instance, or ombrelle d'eau, having at its base a tufty bunch of fine white roots about as thick as a horsehair. In these soft tresses, the caddis worm, which observes a vegetarian diet, will find at one and the same time the wherewithal to build and eat. On the other hand, we have a little faggot of bits of wood, very dry, equal in length and each possessing the thickness of a good sized pin. The two sorts of building material lie side by side, mingling their threads and sticks. The animal can make its choice from the lump. A few hours later, having recovered from the shock of losing its sheath, the caddis worm sets to work to manufacture a new one. It settles across a bunch of tangled rootlets, which are brought together by the builder's legs and more or less arranged by the undulating movement of the hinder part. This gives a kind of incoherent and ill defined suspended belt, a narrow hammock with a number of loose catches; for the various bits of which it is made up are respected by the teeth and extended from place to place beyond the main cords of the roots. Here, without much trouble, is the support, suitably fixed by natural moorings. A few threads of silk, casually distributed, make the frail combination a trifle more secure. And now to the work of building. Supported by the suspended belt, the caddis worm stretches itself and thrusts out its middle legs, which, being longer than the others, are the grapnels intended to seize things at a distance. It meets a bit of root, fastens on to it, climbs above the point gripped, as though it were measuring the piece to a requisite length, and then, with the fine scissors of its mandibles, cuts the string. There is at once a brief recoil, which brings the animal back to the level of the hammock. The bit detached lies across the worm's chest, held in its forelegs, which turn it, twist it, wave it about, lay it down, lift it up, as though trying for the best position. Those forelegs make admirably dexterous arms. Being less long than the other two pairs, they are brought into immediate contact with those primordial implements, the mandibles and the spinneret. Their delicate terminal jointing, with a movable and crooked finger, is the caddis worm's equivalent of our hand. They are the working legs. The second pair, which are exceptionally long, serve to spear distant materials and to give the worker a firm footing when measuring a piece and cutting it with the pliers. Lastly, the hind legs, of medium length, afford a support when the others are busy.