level with the support furnished by the
close tangle of rootlets. With a quick movement, it shifts its burden,
gets it as nearly by the middle as it can, so that the two ends stick
out equally on either side, and chooses the spot to place it, whereupon
the spinneret sets to work at once, while the little fore legs hold the
scrap of root motionless in its transversal position. The soldering
is effected with a touch of silk in the middle of the bit and along a
certain distance to the right and left, as far as the bending of the
head permits.

Without delay, other sticks are speared in like manner at a distance,
cut off and placed in position. As the immediate neighborhood is
stripped, the material is gathered at a yet greater distance and the
caddis worm bends even farther from its support, which now holds only
its last few segments. It is a curious gymnastic display, that of this
soft, hanging spine turning and swaying, while the grapnels feel in
every direction for a thread.

All this labor results in a sort of casing of little white cords.
The work lacks firmness and regularity. Nevertheless, judging by the
builder's methods, I can see that the building would not be devoid
of merit if the materials gave it a better chance. The caddis worm
estimates the size of its pieces very fairly; it cuts them all to nearly
the same length; it always arranges them crosswise on the margin of the
case; it fixes them by the middle.

Nor is this all: the manner of working helps the general arrangement
considerably. When the bricklayer is building the narrow shaft of a
factory chimney, he stands in the center of his turret and turns round
and round while gradually laying new rows. The caddis worm acts in the
same way. It twists round in its sheath; it adopts without inconvenience
whatever position it pleases, so as to bring its spinneret full face
with the point to be gummed. There is no straining of the neck to left
or right, no throwing back of the head to reach points behind.
The animal has constantly before it, within the exact range of its
implements, the place at which the bit is to be fixed. When the piece
is soldered, the worm turns a little aside, to a length equal to that of
the last soldering, and here, along an extent which hardly ever varies,
an extent determined by the swing which its head is able to give, it
fixes the next piece.

These several conditions ought to result in a geometrically ordered
dwelling, having a