I discovered something very strange that has me baffled...after listening to a final mix, I wanted to go back and make 1 subtle tweak (muting guitar strum for 1 measure in middle of song)...all of my tracks were frozen, then bounced to one final mix, then archived so I could work on final mix sweetening without CPU loss. My steps to go back were:
1) archive final mix
2) unarchive other tracks
3) unfreeze guitar track
4) mute already mentioned measure
5) bounce tracks again for final mix #2
When I compared mix #2 to mix #1, I could VISUALLY tell it wasn't as "fat" (amplitude, I guess)...which was also confirmed when I ran it through my mastering software...by almost 1db in places. You may notice that I didn't REFREEZE my guitar track. Sure enough, I did steps 1-4 again...then froze the track...then bounced and it was just like mix #1 again.
What would cause the levels to be different whether or not I bounced frozen/regular tracks? I thought that 2+2 = 4, the same way that 1+1+1+1 = 4...but not in audio rendering perhaps? Fun!