Mats Norrmann:

>>May I suggest that there are two types of rests (Ger. Pause; Fr. silence)
>>- exact and inexact? Exact rests must be strictly observed, inexacts not.
>>Much of it has to do with keeping the overall pulse of the music.  Hence,
>>a rest at the end of a piece or between sections is often not given its
>>full duration, with little harm done.  Needless to say (but I'll probably
>>have to say it anyway, so why not now?), there are also rests that occur
>>in such contexts that should be fully observed.  Performers must actually
>>think about it, rather than simply follow.
>
>Interesting that you are so concerned about observing the rests, but not
>the repeats.

In both cases, the same principle applies: finding out the (or a)
musical reason for the mark.  One difference, however, is that a repeat
is "binary": you take it or you don't.  A rest is "analogue" because
durations are not exact in much of the music played by humans.  If you
have a computer and a MIDI sequencer, try entering a piece like the first
prelude from the Well-Tempered Clavier, and then figure out why it doesn't
sound as good as when Wanda Landowska plays it.

Steve Schwartz