conductor must take care to send them away; since this
large number of human bodies injures the sonority of the instruments. A
symphony performed by an orchestra thus more or less stifled, loses much
of its effect.

There are yet other precautions, relative especially to the orchestra,
which the conductor may also take, to avoid certain defects in performance.
The instruments of percussion, placed, as I have indicated, upon one of
the last rows of the orchestra, have a tendency to modify the rhythm,
and slacken the time. A series of strokes on the drum struck at regular
intervals in a quick movement, like the following:--

[Illustration]

will sometimes lead to the complete destruction of a fine rhythmical
progression, by checking the onward bound of the rest of the orchestra,
and destroying the unity. Almost always, the drum player, through not
observing the original time given by the conductor, is somewhat
behindhand in striking his first stroke. This retardment, multiplied by
the number of strokes which follow the first one, soon produces--as may
be imagined--a rhythmical discrepancy of the most fatal effect. The
conductor,--all whose efforts to re-establish unanimity are then in
vain--has only one thing left to do; which is, to insist that the long
drum player shall count beforehand the number of strokes to be given
in the passage in question, and that, knowing his part, he shall no
longer look at his copy, but keep his eyes constantly fixed upon the
conducting-stick; by which means he will follow the time without the
slightest want of precision.

Another retardment, arising from a different cause, frequently takes
place in the trumpet-parts; it is when they contain a quick flow of
passages such as this:--

[Illustration]

The trumpet-player, instead of taking breath _before_ the first of these
three bars, takes breath at their commencement, during the quaver-rest,
A; and, not counting for anything the short time it has taken him to
breathe, gives its whole value to the quaver-rest, which thus becomes
super-added to the value of the first bar. The result of this is the
following:--

[Illustration]

an effect all the worse because the final accent, struck at the
commencement of the third bar by the rest of the orchestra, comes a
third of the time too slow in the trumpets, and destroys unity in the
striking of the last chord.

To obviate this, the conductor must first previously warn the players
against