Basically, you're doubling the risk of sound quality loss (tripling, even) for like a meg a track space.
Let's put it this way. 0 kbps = no audio data. 8 kbps = original audio data downsampled so much it will sound like garbage.
Why even bother, if you're going to encode the low parts as basically silence/noise?
For a good result, use VBR with a minimum of around 128, and a max at (in my book) 320 kbps. Anything (pretty much) else will generate a low quality file, which is pointless in the age of hundred plus GB harddisks.
It's obvious a proper guide is needed, if we're getting questions like this. And it's not like I know everything (obviously, given some other threads where I proved my ignorance

).
- Spike
Jam it back in, in the dark.