I was looking over my A4 uppers that I have for some upcoming builds and laid a precision straight edge end for end across the top of my uppers to check flatness.  I have a couple CMT  uppers with M4 cuts, one CMT without M4 cuts and one other brand.  One CMT upper shows a 0.004" smile/bow along the upper with the high spots at the end of the upper.  The other two CMT uppers show a 0.007"-0.008" bow in the middle of the uppers.  The other brand upper had less than 0.002" deviation in flatness end for end.  I am not really sure what this information is telling me or if it is really relevant.  I spot checked some other dimensions with a set of calipers and nothing else jumped out as being out of spec.  

What is the tolerance for flatness on the A3/A4 uppers?    I see there is a GD&T callout for parallelism of 0.005", but does that callout govern the flatness of the rail end to end?  Is there a potential problem if the uppers are not completely flat?  How much variation is out there?  It's pretty simple to check with a straight edge and feelers.   I don't want this to turn into a MFG vs. MFG discussion.  

Ultimately, do I have a problem?  Does it matter?  

I believe it has a lot to do with different manufacturers tolerances, they may vary from one manufacturer to the other...

Based on that spec, what is the allowed flatness variation end to end?  I see the .118+.008 and the .005 parallelism callout.  I am trying to understand the 1913 standard a little better(mostly out of curiousity). What are the critical dimensions in the spec that should be looked out for(not just uppers but forends, etc.)?  

Are the uppers out of spec?

Its hard to believe that all of those receivers are out of spec... Although, if they were all purchased from the same person, they very well could be blems...

If I wanted to determine if the upper were out of spec I would first install them level in a milling machine vise that had been 'indictaed in', secure them snugly, and then using the same dial indicator dtermine the true runout from left to right using the indicator. No offense but strainght edges are not always straight. There are also precision ground granite slabs available from firms such as MSC which can be used to determine TRO (total runout). Charles.

As long as you re-attach optics in the same location on the flattop, it really doesn't matter.  I have a colt flattop that is obviously out of spec to the naked eye.  Optics still retain zero if I make sure I put them back on in the same slot they were zeroed in at.

The only surfaces that really matter are the angled edges on either side.  Not aware of any mounts that connect to the upper by making contact with the top surface.