After a couple of minutes in the free air, I replace the caddis worm in the water. This time, it floats, but like a cylinder with too much weight below. The sheath remains vertical, with its hinder orifice level with the water. Soon, an air bubble escapes from the orifice. Deprived of this buoy, the skiff at once goes down.

The result is the same with the caddis worms in shell casings. At first, they float, straight up on end, and then dip under and sink, faster than the others, after sending out an air bubble or two through the back window.

That is enough: the secret is out. When cased in wood or in shells, the caddis worms, which are always heavier than water, are able to keep on the surface by means of a temporary air balloon which decreases the density of the whole structure.

This apparatus works in the simplest manner. Consider the rear of the sheath. It is truncated, wide open and supplied with a membranous partition, the work of the spinneret. A round hole occupies the center of this screen. Beyond it lies the interior of the scabbard, which is smoothly lined and wadded with satin, however rough the exterior may be. Armed at the stern with two hooks which bite into the silky lining, the animal is able to move backwards and forwards at will inside the cylinder, to fix its grapnels at whatever point it pleases and thus to keep a hold on the cylinder while the six legs and the fore part are outside.

When at rest, the body remains indoors entirely and the grub occupies the whole of the tube. But let it contract ever so little towards the front, or, better still, let it stick out a part of its body: a vacuum is formed behind this sort of piston, which may be compared with that of a pump. Thanks to the rear window, a valve without a plug, this vacuum at once fills, thus renewing the aerated water around the gills, a soft fleece of hairs distributed over the back and belly.

The piston stroke affects only the work of breathing; it does not alter the density, makes hardly any change in that which is heavier than water. To lighten the weight, the caddis worm must first rise to the surface. With this object, it scales the grasses of one support after the other; it clambers up, sticking to its purpose in spite of the drawback of its faggot dragging through the tangle.

When it has reached the goal, it lifts the rear end a little above the water and gives a stroke of the piston. The vacuum thus obtained fills with air. That is enough: skiff and boatman are in a position to float. The now useless support of the grasses is abandoned. The time has come for evolutions on the surface, in the glad sunlight.

The caddis worm possesses no great talent as a navigator. To turn round, to tack about, to shift its place slightly by a backward movement is all that it can do; and even that it does very clumsily. The front part of the body, sticking out of the case, acts as a rudder. Three or four times over, it rises abruptly, bends, comes down again and strikes the water. These paddle strokes, repeated at intervals, carry the unskilled oarsman to fresh latitudes. It becomes a voyage on the right seas when the crossing measures a hand's breadth.