NavList:

.25 Stand Dev doesn't sound bad at all.  .14 is strictly hypothetical based on a 20/20 eye and a 7 power scope without any other errors which exist. 

What does come up as odd is the .9 vs apparent zero.  Mine came out so nice.  I can't read mine on the sun any better than your standard deviation with my eyes.  I tested my zero on the same remote objects with the scope with a great match.

I have never done perfect with the sun.  But I cannot imagine filters accounting for such a discrepency. 

For a quick reality check...I would lay it on its side, same place,  set it to zero and shoot same object you zeroed on with the scope...but use your 7 power.....if its on....mmmm  we at least have a start...... if off .9, I suppose would set it up with the scope again and recheck....then leave the sextant on its side, put on the sextant seven power right on the bench and sight on the same object.  That gets rid of filters and the rest of the sun changes.

You can extrapolate the rest of the approach....

Check 7x without filters at identical range as high power scope (same object).

Check closer and farther objects handheld sideways or upright  (roof edge or flagpoles)......look for change in IE  by range. Frank gave us minimum distances in an early post.

Check horizon, then try sun method agian. 

At some point by isolating variables we should figure out exactly when it goes amok, which should tell us where to look for the error.

I did set up one sextant for a friend that was driving him bananas (before I knew how to reduce sights even ). He asked me to look because of my long background in building model plane and boat mechanisms.  He would get it really nice, then it would jump after a little disturbance.  The cure was a replacement mirror tensioner. I would guess it was overstressed at some point so it moved a little too easy when you set it down with a bump or it rode in a car.