You do not know the structure of a MP3-File and how MP3Gain does this, which is explained in the help-file of Mp3Gain:
And to avoid futher misunderstandings:
This "global gain field" is not a tag field, that can we read or written by Mp3Tag. It belongs to the music part of the MP3 and not to the tag part.
There is another method to get gain changes than that done by MP3Gain. Foobar for example is able to store informations in tag fields and during play uses these informations to equalize the music for the hearer. But these informations are only use by Foobar and I think some other compatible players. If you want to apply these changes to be used by all players then you have to do it the way Mp3Gain does. Foobar can do the Mp3Gain-method too in addition to the tag-field method.
So to explain again:
Mp3Gain uses APE-tags to store the gain-informations only for analysis and undo-purpose. These tags are not used by a player to play a a mp3-file louder or less louder. But all players are able to play in the intended loudness because MP3Gain makes changes to the music part of the mp3-file.
There is another method not used by MP3Gain that indeed uses the content of tags after analysis to tell a compatible player, that can read and use these tag-informations, to decide how loud it has to play.
I think you are not aware of these different modes and constantly mix them up.
All depends on the Options/Tags-Settings in MP3Gain and if you tell Mp3Gain not to read and write tags, no undo-information tags are stored.
In addition the presence of APE-Tags written by MP3Gain is no clear evidence, that Mp3Gain has made changes to the global gain field, because MP3Gain writes those tags during analyzing (if writing tags is enabled) and it is possible to tell the program to simply analyze without modifying gain. So there is no real way to filter files with global gain change by filtering the gain-tags in mp3tag.
But of course the presence of these ape-tags is an indication, that probably modifying gain has been done.
If these ape-tags are present you could load these files in MP3Gain (options/tags/recalculate disabled), clear analyzing results (Analyses/Clear Analysis), enable recalculate and analyze again. If the results are the same, modifying gain has been done.
But as you want to modyfy track gain anyway, which is not advisable if you want to keep an album balance, why not modify gain on all mp3s with mp3gain again. I can see no strong need for filtering in advance because the only time needed is for reading these files and not for writing, if modify gain has already been done.
Do you refer to the ReplayGain-method (using tags by a player to play with a defined volume level)?
Or do you mean the method of MP3Gain (using tags only for analalizing and undo-functions but changing the global-gain)?
I thought so after your questions.
If you have treated your files with MP3Gain and you would have understood how Mp3Gain works, you would not have written your questions.
Read again my statement here: Wiping out tag fields
You also should read help-files of the software you use.
Take a look at the help-file of Mp3Gain and there the chapter "Concept" and then it should be clear why it is not a good idea to use software that just "normalizes" as you stated you've done.
was a stupid one, because those values are already added to those MP3s. So it's not an issue
But here is another [hopefully not stupid] question: how do I find among my among my XX XXX MP3 files those, that were treated by MP3Gain at some point?
Given that only recently I have turned on in MP3Gain the "Preserve file date / time" option [and thus starting to preserve the original %_file_mod_date% value] and taking into consideration that I'm after files treated by MP3Gain after 2015 01 01, I thought I could use a filter
"$if($eql(%_file_create_date%,%_file_mod_date%),1,0)" IS "0"and then just look for newer MP3s [in terms of %_file_mod_date%]; but that will get me nowhere. Because something like two weeks ago I have moved all my music to a new drive [and thus making a fresh %_file_create_date% value]