23 navigation stars later...
and here are the results.
The Diff column is after the Altitude correction is applied. I have noticed that all the differences are positive. Usually an accurate sextant would give small errors distributed around zero, but ours is distributed around 13'. This qualifies as an index correction that can be subtracted from all measurements with this sextant to make it more accurate. Real sextants have index measurements arising from their mechanical construction. I'm guessing mine might be a function of my calibration system, sextant construction, and measurement bias when estimating Hs. Your mileage might vary, but I will consider using it going forward.
I have included a sample calculation that shows how to apply the Index correction if you choose to do so. Note that the average error is now less than 5' or 10 km. Well within my specification of 30 km so far. Time for something to go horribly wrong.
