Oh... and of course once you get whatever input gain automation you need done you just bounce it so it's printed right on the track and you don't have to think about it anymore. If you put you meter in the PC chain like I described any Gain adjustments you make will be reflected in the meter so if you want to see a before and after of THAT clone the original (which will remain untouched) and do your changes on the clone. So that way if your hunting for original spikes reference the original track. Do all your changes in the clone.
 
Or something like that.
 
If there is like a TON of spikes and tons of clips or whatever, honestly it might just be better to widdle away at them with a limiter/compressor (to tame the spikes) then send that through your effects. I was working on the assumption that it was a manageable number of screwy peaks but if you gotta automate like HUNDREDS of them on each track... I'd just use a limiter/compressor to save time if it's not sooooper important.
 

Yeah - thanks for all the thinking on this - I had a really brutal day with multiple doctor appointments, and I am going to lie down and see if my spine will stop yelling at me.  I'll read back through the thread a bit later and see what I can do with it all.  And, yup there are loads of takes throughout.
 
I'll get it - and thanks again,
 
Bob Bone
 

If you have the free version of AA3, I believe that also has batch algorithm capability (I use AA4). You can batch normalize audio files in a few seconds with it.
 
Quick Edit: If you choose this route, just in case these are takes embedded into a track wav file, you may have to go the route of "Save As..." and check "Copy all audio with project" as well as "Create one file per clip." This would get messy with the clip count you mentioned, but with takes this can also be done by pulling take lane data into individual tracks so they are discrete wav files. A batch conversion process needs to see each as an individual wav (either take or clip, depending).

Edit 2: Looks like the new clip export (Ipswich) will allow you to timestamp clips so will definitely ease this method.

What I would do is, route the output to a bus, switch on the waveform preview for that bus in track view, play through the track, and thereby get a broad visual representation of the relative levels from one clip to the next. Drag the decibel scale in the space between the bus header and the bus itself to make sure you can clearly see the waveform size differences, and maybe even maximise the bus to make it easier to see. Pick the 5 or 10 clips that look to be loudest, drop each of their gain automation down by 3dB. Repeat. Eventually they will be broadly in line and you might want to nudge one or two up or down a dB or two. Then adjust the track gain to taste.
 
I don't think you need to worry about RMS or compression etc. If the only difference between the takes is the input level, then scaling the gain will bring everything into line (signal/noise ratio aside, but that's unfixable, but also thankfully unlikely to be relevant). If however there are other differences between the performances, such as the guitarist using a different signal chain in each case, that will require a more considered approach.

getting peaks consistent,
boils down to individual tracking, track per track.
 
your Modus Operendi should be for each track to be in it's 'sweet spot',
then they should all be about the same, peak wise...
you do this with gain structure,
mic placement,
dynamic effects, 
everything that is 'recording'.
 
this is something that should happen during tracking,
not after.
 

Thanks for the input - @Batsbrew - I completely agree.  The guitar player recorded a bunch of tracks off by himself, and has no recording experience, so he didn't know to match levels up front.
 
Bob Bone