les the animal with a stick, till it looses its hold, and
allows itself to be taken without further resistance. At other times,
when pursued, and finding flight ineffectual, the Armadillos withdraw
the head under the edge of the buckler of the shoulders; their legs,
except the feet, are naturally hidden by the borders of the bucklers and
the bands; they then contract the body as far towards the shape of a
ball as the stretching of the membrane which unites the different
movable pieces of the armour will permit.[8] Thus defended, they
frequently escape danger; but if near a precipice, the animal will
sometimes roll itself over, and in this case, says Molina, in his
Natural History of Chili, it generally falls to the bottom unhurt.

[8] It should here be observed that the Three-banded Armadillo
is remarkable for the faculty of rolling itself up more
completely than the other species. It can, in so doing, totally
conceal the head, the tail, and the fore feet, which none of the
other species can completely effect.--_Cuvier._

Armadillos were formerly thought to feed exclusively on vegetables; but
they have since been found to devour insects and flesh. The directions
of their burrows evince that they search after ant heaps, and the
insects quickly disappear from near the hole of an Armadillo. The
largest species, the great black Armadillo, common in the forests of
Paraguay, feeds on the carcasses of animals; and the graves of the dead
which are necessarily formed at a distance from the usual places of
sepulture, in countries where the great Armadillo is found, are
protected by strong double boards to prevent the animal from penetrating
and devouring the body. It appears, also, that it eats young birds,
eggs, snakes, lizards, &c. The Indians are very fond of the flesh of the
Armadillo as food, especially when young; but, when old, it acquires a
strong musky flavour. Mr. Waterton, who tasted the flesh, considered it
strong and rank. The shells or crusts are applied to various useful
purposes, and painted of different colours are made into boxes, baskets,
&c.

Cuvier remarks that that old mode of distinguishing the species of
Armadillos by the number of the bands is clearly objectionable, inasmuch
as D'Azara has established that not only the number of these bands
varies, in the different individuals of the same species, but further,
that there are individuals of different species which have the same
numbe