the sound from a real watchman must necessarily have entered by the
window, the difference between the two directions was considerably less
than that which the ear is unable to appreciate. Had the ventriloquist
stood at one window, and tried to make the sound of a watchman's voice
enter _another_ window, he would have failed in his performance,
because the difference of the two directions was too great. In like
manner, when M. Alexandre introduced a boy from the street, and made him
sing from his stomach the song of Malbrook, he placed his head as near as
possible to the boy's chest, under the pretence of listening, whereas the
real object of it was to assimilate as much as possible the true and the
fictitious direction of the sounds. Had he placed the boy at the distance
of six or eight feet, the real singer would have been soon detected.

We have made several experiments with a view of determining the angle of
uncertainty, or the angle within which the ear cannot discover the
direction of sounds; but this is not easily done, for it varies with the
state of the air and of surrounding objects. If the air is perfectly pure,
and if no objects surround the sounding body, the angle of uncertainty
will be less than under any other circumstances, as the sound suffers
neither deviation nor reflection. If the sounding body is encircled with
objects which reflect sound, the echoes arrive at the ear, at short
distances, nearly at the same time with the direct sound; and as they form
a single sound, the angle of uncertainty must then be much greater, for
the sound really arrives at the ear from various quarters. The
ventriloquist, therefore, might avail himself of this principle, and
choose an apartment in which the reverberations from its different sides
multiply the directions of the sounds which he utters, and thus facilitate
his purpose of directing the imagination of his audience to the object
from which he wishes these sounds to be thought to proceed.