As a newbie in relation to both fb2000 and ReplayGain, I find what follows to be a bit of a puzzle... When I look at the results of an RG scan, I can see that, for most albums, all the album peak values displayed against the tracks of a given album are equal, and so are all the album gain values. This is exactly what I would expect if I understand RG correctly. However, there are a few albums for which this does not seem to apply. For such an album, some or all of the tracks within it yield different album peak values and correspondingly different album gain values. Does this make any sense? Surely an album can have only one overall peak and only one average gain applied?
If you used the "Scan selection as albums (by tags)" ReplayGain method it is possible to get differing values if the tags don't match. The scanner by default uses %album artist%, %date% and %album% tags to group albums together so a difference in any of these three fields would result the track to be treated as being from a different album.
Once you fix any differences in these tags I'm sure the album gain values will get consistent. Or if necessary, you can customize the grouping pattern in Preferences -> Advanced -> Tools -> ReplayGain Scanner -> Album grouping pattern. I personally use a stricter separation and have %discnumber% in the grouping pattern too so each disc in a multi-disc set has its own gain value.
Thanks Case - I think you've hit the nail right on the head!